From ahmed.bhayat at manchester.ac.uk Sun Jan 6 06:54:56 2019 From: ahmed.bhayat at manchester.ac.uk (Ahmed Bhayat) Date: Sun, 6 Jan 2019 11:54:56 +0000 Subject: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction Message-ID: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> Dear all, Would welcome information on the following: Does there exist a reduction ordering, total on ground terms of the simply-typed lambda calculus, that is compatible with beta-reduction? That is, does there exist a reduction order ?>?, total on ground terms of STT, such that for all terms t1 and t2 such that t1 beta-reduces to t2, we have t1 > t2? Alternatively, is there a proof that no such ordering can exist? Failing either of the above, I would be interested in any partial results or possibly relevant results. Thanks Ahmed Bhayat ahmed.bhayat at manchester.ac.uk PhD candidate at the University of Manchester From jamesbkoppel at yahoo.com Sun Jan 6 18:25:07 2019 From: jamesbkoppel at yahoo.com (James Koppel) Date: Sun, 6 Jan 2019 18:25:07 -0500 Subject: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction In-Reply-To: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> References: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> Message-ID: Hello Ahmed, Because the STLC is strongly-normalizing, beta reduction is itself a reduction order. Terms A and B which do not transitively reduce to each other may then be ordered arbitrarily. I believe this satisfies everything you're looking for. Yours, Jimmy Koppel MIT On Sun, Jan 6, 2019 at 5:57 PM Ahmed Bhayat wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear all, > > Would welcome information on the following: > > Does there exist a reduction ordering, total on ground terms of the > simply-typed lambda calculus, that is compatible with beta-reduction? > > That is, does there exist a reduction order ?>?, total on ground terms of > STT, such that for all terms t1 and t2 such that t1 beta-reduces to t2, we > have t1 > t2? > > Alternatively, is there a proof that no such ordering can exist? > > Failing either of the above, I would be interested in any partial results > or possibly relevant results. > > Thanks > > Ahmed Bhayat > > ahmed.bhayat at manchester.ac.uk > PhD candidate at the University of Manchester > From Damiano.Mazza at lipn.univ-paris13.fr Mon Jan 7 03:18:27 2019 From: Damiano.Mazza at lipn.univ-paris13.fr (Damiano Mazza) Date: Mon, 7 Jan 2019 09:18:27 +0100 Subject: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction In-Reply-To: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> References: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> Message-ID: <43c22207-97c1-1e3d-3d81-40c548e995db@lipn.univ-paris13.fr> Dear Ahmed, I'm not sure I fully understand what you are asking, if you're looking for a total order on the set of simply-typed lambda-terms such that t --> t' implies t < t', then you can do as follows. In the simply-typed lambda-calculus, every term strongly normalizes, so you may define an integer #t which is equal to the length of the longest reduction of t to its normal form. This induces a partition on the set of simply-typed lambda-terms in countably many sets, call them S_n, such that S_n contains all terms t s.t. #t = n. Each S_n is of course countable, so you may pick an injection beta_n : S_n ---> N, which induces a total ordering within S_n. Observe now that t --> t' implies #t' < #t, which means that no two terms in the same S_n beta-reduce to one another. Therefore, the map t |---> (#t, beta_n(t)). induces a total order on the simply-typed lambda-terms isomorphic to omega^2 with the desired property. Best wishes, Damiano On 06/01/2019 12:54, Ahmed Bhayat wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear all, > > Would welcome information on the following: > > Does there exist a reduction ordering, total on ground terms of the simply-typed lambda calculus, that is compatible with beta-reduction? > > That is, does there exist a reduction order ?>?, total on ground terms of STT, such that for all terms t1 and t2 such that t1 beta-reduces to t2, we have t1 > t2? > > Alternatively, is there a proof that no such ordering can exist? > > Failing either of the above, I would be interested in any partial results or possibly relevant results. > > Thanks > > Ahmed Bhayat > > ahmed.bhayat at manchester.ac.uk > PhD candidate at the University of Manchester > From frederic.blanqui at inria.fr Mon Jan 7 03:55:41 2019 From: frederic.blanqui at inria.fr (=?UTF-8?B?RnLDqWTDqXJpYyBCbGFucXVp?=) Date: Mon, 7 Jan 2019 09:55:41 +0100 Subject: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction In-Reply-To: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> References: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> Message-ID: <889b3012-d6c4-43fa-f052-b18bfe79fa9b@inria.fr> Hello. I am not aware of such results. Concerning reduction orders on simply-typed lambda-terms, you can have a look at the works of Jouannaud and Rubio: http://doi.org/10.1145/1206035.1206037 and http://doi.org/10.1145/2699913. Best regards, Fr?d?ric. Le 06/01/2019 ? 12:54, Ahmed Bhayat a ?crit?: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear all, > > Would welcome information on the following: > > Does there exist a reduction ordering, total on ground terms of the simply-typed lambda calculus, that is compatible with beta-reduction? > > That is, does there exist a reduction order ?>?, total on ground terms of STT, such that for all terms t1 and t2 such that t1 beta-reduces to t2, we have t1 > t2? > > Alternatively, is there a proof that no such ordering can exist? > > Failing either of the above, I would be interested in any partial results or possibly relevant results. > > Thanks > > Ahmed Bhayat > > ahmed.bhayat at manchester.ac.uk > PhD candidate at the University of Manchester From ahmed.bhayat at manchester.ac.uk Mon Jan 7 06:02:46 2019 From: ahmed.bhayat at manchester.ac.uk (Ahmed Bhayat) Date: Mon, 7 Jan 2019 11:02:46 +0000 Subject: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction In-Reply-To: References: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk>, Message-ID: <9428AF879CF6F248822A10EFD17423BBDD269B@MBXP11.ds.man.ac.uk> Thank you James, It does not appear to meet the requirements. As a reduction ordering is being sought, compatibility with contexts is required. Consider the following terms: 1) t1 = ^x. (f x x) and t2 = (f a). As we do not have that t1 -->beta t2 or t2 -->beta t1, an arbitrary decision must be made as to the ordering. Choose t2 > t1. But, then in the context []a, [^x. (f x x)]a -->beta [fa]a and thus by compatibility with beta-reduction [t1]a > [t2]a contradicting compatibility with contexts. Further, as a reduction ordering is sought, the ordering must be well-founded. Making an arbitrary decision does not ensure this. Consider terms: 2) t1 = (^x. f x x x x)t and some arbitrary term t2. By compatibility with beta-reduction we must have t1 > f t t t t. We then make an arbitrary choice and have f t t t t > t2. If by another arbitrary choice t2 > t1, well-founded has clearly been lost. Regards, Ahmed ________________________________ From: James Koppel [jkoppel at mit.edu] Sent: Sunday, January 06, 2019 11:22 PM To: Ahmed Bhayat Cc: types-list at lists.seas.upenn.edu Subject: Re: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction Hello Ahmed, Because the STLC is strongly-normalizing, beta reduction is itself a reduction order. Terms A and B which do not transitively reduce to each other may then be ordered arbitrarily. I believe this satisfies everything you're looking for. Yours, Jimmy Koppel MIT On Sun, Jan 6, 2019 at 5:57 PM Ahmed Bhayat > wrote: [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] Dear all, Would welcome information on the following: Does there exist a reduction ordering, total on ground terms of the simply-typed lambda calculus, that is compatible with beta-reduction? That is, does there exist a reduction order ?>?, total on ground terms of STT, such that for all terms t1 and t2 such that t1 beta-reduces to t2, we have t1 > t2? Alternatively, is there a proof that no such ordering can exist? Failing either of the above, I would be interested in any partial results or possibly relevant results. Thanks Ahmed Bhayat ahmed.bhayat at manchester.ac.uk PhD candidate at the University of Manchester -- James Koppel twitter.com/jimmykoppel www.jameskoppelcoaching.com From jamesbkoppel at yahoo.com Mon Jan 7 17:05:40 2019 From: jamesbkoppel at yahoo.com (James Koppel) Date: Mon, 7 Jan 2019 17:05:40 -0500 Subject: [TYPES] Reduction ordering on terms of simply-typed lambda-calculus compatible with beta-reduction In-Reply-To: <9428AF879CF6F248822A10EFD17423BBDD269B@MBXP11.ds.man.ac.uk> References: <9428AF879CF6F248822A10EFD17423BBDD2645@MBXP11.ds.man.ac.uk> <9428AF879CF6F248822A10EFD17423BBDD269B@MBXP11.ds.man.ac.uk> Message-ID: Thanks for the explanation, Ahmed. I confess to not having studied reduction orderings in higher-order term rewriting, only in first-order term-rewriting. Problem 2 still seems easy to circumvent, however. On Mon, Jan 7, 2019 at 7:38 AM Ahmed Bhayat wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Thank you James, > > It does not appear to meet the requirements. As a reduction ordering is > being sought, compatibility with contexts is required. Consider the > following terms: > > 1) t1 = ^x. (f x x) and t2 = (f a). As we do not have that t1 -->beta t2 > or t2 -->beta t1, an arbitrary decision must be made as to the ordering. > Choose t2 > t1. But, then in the context []a, [^x. (f x x)]a -->beta [fa]a > and thus by compatibility with beta-reduction [t1]a > [t2]a contradicting > compatibility with contexts. > > Further, as a reduction ordering is sought, the ordering must be > well-founded. Making an arbitrary decision does not ensure this. Consider > terms: > 2) t1 = (^x. f x x x x)t and some arbitrary term t2. By compatibility with > beta-reduction we must have t1 > f t t t t. We then make an arbitrary > choice and have f t t t t > t2. If by another arbitrary choice t2 > t1, > well-founded has clearly been lost. > > Regards, > > Ahmed > ________________________________ > From: James Koppel [jkoppel at mit.edu] > Sent: Sunday, January 06, 2019 11:22 PM > To: Ahmed Bhayat > Cc: types-list at lists.seas.upenn.edu > Subject: Re: [TYPES] Reduction ordering on terms of simply-typed > lambda-calculus compatible with beta-reduction > > Hello Ahmed, > > Because the STLC is strongly-normalizing, beta reduction is itself a > reduction order. Terms A and B which do not transitively reduce to each > other may then be ordered arbitrarily. I believe this satisfies everything > you're looking for. > > Yours, > Jimmy Koppel > MIT > > On Sun, Jan 6, 2019 at 5:57 PM Ahmed Bhayat > wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear all, > > Would welcome information on the following: > > Does there exist a reduction ordering, total on ground terms of the > simply-typed lambda calculus, that is compatible with beta-reduction? > > That is, does there exist a reduction order ?>?, total on ground terms of > STT, such that for all terms t1 and t2 such that t1 beta-reduces to t2, we > have t1 > t2? > > Alternatively, is there a proof that no such ordering can exist? > > Failing either of the above, I would be interested in any partial results > or possibly relevant results. > > Thanks > > Ahmed Bhayat > > ahmed.bhayat at manchester.ac.uk > PhD candidate at the University of Manchester > > > -- > James Koppel > twitter.com/jimmykoppel > www.jameskoppelcoaching.com > From cwjnkins at gmail.com Mon Jan 7 21:09:39 2019 From: cwjnkins at gmail.com (Christopher Jenkins) Date: Mon, 7 Jan 2019 20:09:39 -0600 Subject: [TYPES] Literature relating subtyping to positivity checking for recursive types? Message-ID: In Curry-style type systems, it is possible for a term to have two different types; in the setting that I am interested in (Cedille) it is possible to define a type of "Cast S T" witnessing that all terms of type S also have type T, which in turn induces a form of subtyping S <: T. I would like to know whether anyone knows about or can point me to work related to the following two subjects: 1. type systems in which the subtyping judgment is replaced by a type (like Cast) and where subtyping inference rules elaborate terms with inserted coercions (Luo's "Coercive Subtyping" is closely related in this second respect) 2. the connection between subtyping and positivity checking for recursive types. To expand on this, let F be some type-scheme of kind Type -> Type. Often (such as when working with inductive data-types as least fixpoints of functors) one is interested in restricting F to be positive. With Cast as the notion of subtyping, this requirement can be stated as the existence of a proof / function of type X <: Y -> F X <: F Y, in effect allowing one to replace the usual detailed account of positivity checking with a set of (elaborating) subtyping rules and saying "F is positive iff F X <: F Y is derivable assuming X <: Y is in the context". Maybe this second observation isn't very deep, but nonetheless I could not find any work reducing positivity checking to a subtyping problem, and wanted to know if someone else has also made this connection. Thanks, From gabriel.scherer at gmail.com Tue Jan 8 05:26:24 2019 From: gabriel.scherer at gmail.com (Gabriel Scherer) Date: Tue, 8 Jan 2019 11:26:24 +0100 Subject: [TYPES] Literature relating subtyping to positivity checking for recursive types? In-Reply-To: References: Message-ID: Dear Christopher, I would say that the observations you have made hold in many systems (and are well-understood), but that their usefulness, in reducing one problem into the other, highly depends on the pragmatics of the specific system you are working on. In most cases, my intuition is that the *converse* transformation to the one you highlight is more convenient, so it is an interesting observation if you can demonstrate that Cedille goes against the grain on that topic -- as well. 1. (Cast S T) can be defined/derived in many systems. For example, if you have definitional subtyping, (t : S <: T) being the coercion of (t : S) into a T when (S <: T) holds, and a propositional equality, you can define Cast S T := Sigma (f : S -> T). (f = lambda (x : S). (x : S <: T)) I would expect that, in most system, using a subtyping judgment directly is more convenient that searching for inhabitants of Cast. NuPRL (where I expect that you can just write (f = lam x. x)) might be a place to look for the converse phenomenon. 2. I think that the direct relation between fixpoints and positivity (and monotonicity) is well-known, but then again, I expect that in most cases it is simpler to establish a property such as (F : * -{+}-> *) by reasoning on the structure of F than a relational property such as (A <: B). For an example of direct connection between positivity-in-fixpoints and positivity-for-subtyping, see the relation between Fixed Points of Type Constructors and Primitive Recursion Abel and Mattes, 2004 and Polarized Subtyping for Sized Types Abel, 2006 The latter has a system for subtyping in F-omega (arrow kinds (k -> k') are extended to have a variance (k -{v}-> k'), typically v \in {+, -}) which directly extends the former's system to check positivity for fixpoint types. I think it makes sense to call (F : k) the "structural" approach, and (F X <: F Y) the "relational" approach to establish positivity. Most developments of the "structural" approach and its meta-theory contain a proof that structural judgments imply relational judgments: if you prove (F : * -{+} -> *), then for any (X <: Y) you have (F X <: F Y). (In the polarized presentation above the presentation is slightly different and I like it: if you prove (G |- A : +), then (g <: g' : G) implies (A[g] <: A[g'])). Conversely, many works on subtyping prove inversion lemmas on the subtyping judgment (often used to define an "algorithmic" subtyping judgment), typically that (A -> B <: A' -> B') holds iff (A :> A') and (B <: B'). For systems where all connectives are invertible in this way, you know that the structural approch is complete wrt. the relational approach: to prove a goal of the form (F X <: F Y), you can reason only on the structure of F by repeated inversion, all subgoals remains of the form (T[X] <: T[Y]) until the axiom rules, and the resulting relational derivation is isomorphic to a structural derivation (only on F). The relational approach would be more interesting in a system where non-structural rules are more expressive (they are not admissible in the structural fragment by cut-elimination or such): where some subtyping judgments start with two types of the same shape, but their premises have different shapes. I don't know if if it is the case in your system. Finally (and slightly off-topic), there are some cases where a relational judgment of the form (G |- A <: B) can be seen as a non-relational type system, by seeing (<:) as a connective with an introduction rule (into something like (G; A |- B). There is such a presentation of subtyping with intersection types as the "ISC system", a sequent calculus for deriving intersection subtypings, in Intersection Subtyping with Constructors Olivier Laurent, 2018 On Tue, Jan 8, 2019 at 9:55 AM Christopher Jenkins wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > In Curry-style type systems, it is possible for a term to have two > different types; in the setting that I am interested in (Cedille) it is > possible to define a type of "Cast S T" witnessing that all terms of type S > also have type T, which in turn induces a form of subtyping S <: T. > > I would like to know whether anyone knows about or can point me to work > related to the following two subjects: > > 1. type systems in which the subtyping judgment is replaced by a type (like > Cast) and where subtyping inference rules elaborate terms with inserted > coercions (Luo's "Coercive Subtyping" is closely related in this second > respect) > > 2. the connection between subtyping and positivity checking for recursive > types. To expand on this, let F be some type-scheme of kind Type -> Type. > Often (such as when working with inductive data-types as least fixpoints of > functors) one is interested in restricting F to be positive. With Cast as > the notion of subtyping, this requirement can be stated as the existence of > a proof / function of type X <: Y -> F X <: F Y, in effect allowing one to > replace the usual detailed account of positivity checking with a set of > (elaborating) subtyping rules and saying "F is positive iff F X <: F Y is > derivable assuming X <: Y is in the context". > > Maybe this second observation isn't very deep, but nonetheless I could not > find any work reducing positivity checking to a subtyping problem, and > wanted to know if someone else has also made this connection. > > Thanks, > From sandro.stucki at gmail.com Tue Jan 8 06:32:22 2019 From: sandro.stucki at gmail.com (Sandro Stucki) Date: Tue, 8 Jan 2019 12:32:22 +0100 Subject: [TYPES] Literature relating subtyping to positivity checking for recursive types? In-Reply-To: References: Message-ID: Dear Christopher, In addition to the work by Abel and others mentioned by Gabriel, I'd like to point out Martin Steffen's thesis Polarized Higher-Order Subtyping Martin Steffen, PhD thesis, 1998 https://heim.ifi.uio.no/msteffen/download/diss/diss.pdf which starts out with a presentation of polarized subtyping that has explicit kinding rules of the form "?X.T has kind k1 -{+}-> k2 iff T[Y/X] <: T[Z/X]: k2 is derivable assuming Y <: Z: k1 is in the context" very similar to the one you suggest in your email (see the various K-Arrow-I rules in ch. 4.4). To establish decidability of subtyping, Steffen then goes on to show that this system is equivalent to one more closely resembling Abel's (following the "structural" approach mentioned by Gabriel). That latter system makes kinding independent of subtyping by introducing an explicit judgment for positivity checking (or rather, variance checking) instead. Best regards /Sandro On Tue, Jan 8, 2019 at 11:20 AM Gabriel Scherer wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear Christopher, > > I would say that the observations you have made hold in many systems > (and are well-understood), but that their usefulness, in reducing one > problem into the other, highly depends on the pragmatics of the > specific system you are working on. In most cases, my intuition is > that the *converse* transformation to the one you highlight is more > convenient, so it is an interesting observation if you can demonstrate > that Cedille goes against the grain on that topic -- as well. > > 1. (Cast S T) can be defined/derived in many systems. For example, if > you have definitional subtyping, (t : S <: T) being the coercion of > (t : S) into a T when (S <: T) holds, and a propositional equality, > you can define > > Cast S T := Sigma (f : S -> T). (f = lambda (x : S). (x : S <: T)) > > I would expect that, in most system, using a subtyping judgment > directly is more convenient that searching for inhabitants of > Cast. NuPRL (where I expect that you can just write (f = lam x. x)) > might be a place to look for the converse phenomenon. > > > 2. I think that the direct relation between fixpoints and positivity > (and monotonicity) is well-known, but then again, I expect that in > most cases it is simpler to establish a property such as > (F : * -{+}-> *) by reasoning on the structure of F than a relational > property such as (A <: B). > > For an example of direct connection between positivity-in-fixpoints > and positivity-for-subtyping, see the relation between > > Fixed Points of Type Constructors and Primitive Recursion > Abel and Mattes, 2004 > > and > > Polarized Subtyping for Sized Types > Abel, 2006 > > The latter has a system for subtyping in F-omega (arrow kinds (k -> > k') are extended to have a variance (k -{v}-> k'), typically v \in > {+, -}) which directly extends the former's system to check positivity > for fixpoint types. > > I think it makes sense to call (F : k) the "structural" approach, and > (F X <: F Y) the "relational" approach to establish positivity. > > Most developments of the "structural" approach and its meta-theory > contain a proof that structural judgments imply relational judgments: > if you prove (F : * -{+} -> *), then for any (X <: Y) you have (F X <: > F Y). (In the polarized presentation above the presentation is > slightly different and I like it: if you prove (G |- A : +), then (g > <: g' : G) implies (A[g] <: A[g'])). > > Conversely, many works on subtyping prove inversion lemmas on the > subtyping judgment (often used to define an "algorithmic" subtyping > judgment), typically that (A -> B <: A' -> B') holds iff (A :> A') and > (B <: B'). For systems where all connectives are invertible in this > way, you know that the structural approch is complete wrt. the > relational approach: to prove a goal of the form (F X <: F Y), you can > reason only on the structure of F by repeated inversion, all subgoals > remains of the form (T[X] <: T[Y]) until the axiom rules, and the > resulting relational derivation is isomorphic to a structural > derivation (only on F). > > The relational approach would be more interesting in a system where > non-structural rules are more expressive (they are not admissible in > the structural fragment by cut-elimination or such): where some > subtyping judgments start with two types of the same shape, but their > premises have different shapes. I don't know if if it is the case in > your system. > > Finally (and slightly off-topic), there are some cases where > a relational judgment of the form (G |- A <: B) can be seen as > a non-relational type system, by seeing (<:) as a connective with an > introduction rule (into something like (G; A |- B). There is such > a presentation of subtyping with intersection types as the "ISC > system", a sequent calculus for deriving intersection subtypings, in > > Intersection Subtyping with Constructors > Olivier Laurent, 2018 > > > On Tue, Jan 8, 2019 at 9:55 AM Christopher Jenkins > wrote: > > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > > ] > > > > In Curry-style type systems, it is possible for a term to have two > > different types; in the setting that I am interested in (Cedille) it is > > possible to define a type of "Cast S T" witnessing that all terms of type S > > also have type T, which in turn induces a form of subtyping S <: T. > > > > I would like to know whether anyone knows about or can point me to work > > related to the following two subjects: > > > > 1. type systems in which the subtyping judgment is replaced by a type (like > > Cast) and where subtyping inference rules elaborate terms with inserted > > coercions (Luo's "Coercive Subtyping" is closely related in this second > > respect) > > > > 2. the connection between subtyping and positivity checking for recursive > > types. To expand on this, let F be some type-scheme of kind Type -> Type. > > Often (such as when working with inductive data-types as least fixpoints of > > functors) one is interested in restricting F to be positive. With Cast as > > the notion of subtyping, this requirement can be stated as the existence of > > a proof / function of type X <: Y -> F X <: F Y, in effect allowing one to > > replace the usual detailed account of positivity checking with a set of > > (elaborating) subtyping rules and saying "F is positive iff F X <: F Y is > > derivable assuming X <: Y is in the context". > > > > Maybe this second observation isn't very deep, but nonetheless I could not > > find any work reducing positivity checking to a subtyping problem, and > > wanted to know if someone else has also made this connection. > > > > Thanks, > > From cwjnkins at gmail.com Tue Jan 8 15:22:46 2019 From: cwjnkins at gmail.com (Christopher Jenkins) Date: Tue, 8 Jan 2019 14:22:46 -0600 Subject: [TYPES] Literature relating subtyping to positivity checking for recursive types? In-Reply-To: References: Message-ID: Hi, Gabriel and Sandro, Thank you very much for the pointers and information! This was very helpful, especially so the structural / relational distinction for approaches to subtyping. Gabriel, 1. (Cast S T) can be defined/derived in many systems. For example, if > you have definitional subtyping, (t : S <: T) being the coercion of > (t : S) into a T when (S <: T) holds, and a propositional equality, > you can define > > Cast S T := Sigma (f : S -> T). (f = lambda (x : S). (x : S <: T)) > > I would expect that, in most system, using a subtyping judgment > directly is more convenient that searching for inhabitants of > Cast. NuPRL (where I expect that you can just write (f = lam x. x)) > might be a place to look for the converse phenomenon. > What you have described as the NuPRL approach is exactly what we use to define Cast in Cedille. The only notion of subtyping in the core theory is the one observed by Miquel for the Implicit Calculus of Constructions as typing judgments of the form G,x:S |- x:T, and I'm interested in the design space of languages with subtyping derived from this. Sandro, which starts out with a presentation of polarized subtyping that has > explicit kinding rules of the form "?X.T has kind k1 -{+}-> k2 iff > T[Y/X] <: T[Z/X]: k2 is > derivable assuming Y <: Z: k1 is in the context" very similar to the > one you suggest in your email > Excellent, this sounds like what I was looking for! Thanks again, On Tue, Jan 8, 2019 at 5:32 AM Sandro Stucki wrote: > Dear Christopher, > > In addition to the work by Abel and others mentioned by Gabriel, I'd > like to point out Martin Steffen's thesis > > Polarized Higher-Order Subtyping > Martin Steffen, PhD thesis, 1998 > https://heim.ifi.uio.no/msteffen/download/diss/diss.pdf > > which starts out with a presentation of polarized subtyping that has > explicit kinding rules of the form "?X.T has kind k1 -{+}-> k2 iff > T[Y/X] <: T[Z/X]: k2 is > derivable assuming Y <: Z: k1 is in the context" very similar to the > one you suggest in your email (see the various K-Arrow-I rules in ch. > 4.4). To establish decidability of subtyping, Steffen then goes on to > show that this system is equivalent to one more closely resembling > Abel's (following the "structural" approach mentioned by Gabriel). > That latter system makes kinding independent of subtyping by > introducing an explicit judgment for positivity checking (or rather, > variance checking) instead. > > Best regards > /Sandro > > > On Tue, Jan 8, 2019 at 11:20 AM Gabriel Scherer > wrote: > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > Dear Christopher, > > > > I would say that the observations you have made hold in many systems > > (and are well-understood), but that their usefulness, in reducing one > > problem into the other, highly depends on the pragmatics of the > > specific system you are working on. In most cases, my intuition is > > that the *converse* transformation to the one you highlight is more > > convenient, so it is an interesting observation if you can demonstrate > > that Cedille goes against the grain on that topic -- as well. > > > > 1. (Cast S T) can be defined/derived in many systems. For example, if > > you have definitional subtyping, (t : S <: T) being the coercion of > > (t : S) into a T when (S <: T) holds, and a propositional equality, > > you can define > > > > Cast S T := Sigma (f : S -> T). (f = lambda (x : S). (x : S <: T)) > > > > I would expect that, in most system, using a subtyping judgment > > directly is more convenient that searching for inhabitants of > > Cast. NuPRL (where I expect that you can just write (f = lam x. x)) > > might be a place to look for the converse phenomenon. > > > > > > 2. I think that the direct relation between fixpoints and positivity > > (and monotonicity) is well-known, but then again, I expect that in > > most cases it is simpler to establish a property such as > > (F : * -{+}-> *) by reasoning on the structure of F than a relational > > property such as (A <: B). > > > > For an example of direct connection between positivity-in-fixpoints > > and positivity-for-subtyping, see the relation between > > > > Fixed Points of Type Constructors and Primitive Recursion > > Abel and Mattes, 2004 > > > > and > > > > Polarized Subtyping for Sized Types > > Abel, 2006 > > > > The latter has a system for subtyping in F-omega (arrow kinds (k -> > > k') are extended to have a variance (k -{v}-> k'), typically v \in > > {+, -}) which directly extends the former's system to check positivity > > for fixpoint types. > > > > I think it makes sense to call (F : k) the "structural" approach, and > > (F X <: F Y) the "relational" approach to establish positivity. > > > > Most developments of the "structural" approach and its meta-theory > > contain a proof that structural judgments imply relational judgments: > > if you prove (F : * -{+} -> *), then for any (X <: Y) you have (F X <: > > F Y). (In the polarized presentation above the presentation is > > slightly different and I like it: if you prove (G |- A : +), then (g > > <: g' : G) implies (A[g] <: A[g'])). > > > > Conversely, many works on subtyping prove inversion lemmas on the > > subtyping judgment (often used to define an "algorithmic" subtyping > > judgment), typically that (A -> B <: A' -> B') holds iff (A :> A') and > > (B <: B'). For systems where all connectives are invertible in this > > way, you know that the structural approch is complete wrt. the > > relational approach: to prove a goal of the form (F X <: F Y), you can > > reason only on the structure of F by repeated inversion, all subgoals > > remains of the form (T[X] <: T[Y]) until the axiom rules, and the > > resulting relational derivation is isomorphic to a structural > > derivation (only on F). > > > > The relational approach would be more interesting in a system where > > non-structural rules are more expressive (they are not admissible in > > the structural fragment by cut-elimination or such): where some > > subtyping judgments start with two types of the same shape, but their > > premises have different shapes. I don't know if if it is the case in > > your system. > > > > Finally (and slightly off-topic), there are some cases where > > a relational judgment of the form (G |- A <: B) can be seen as > > a non-relational type system, by seeing (<:) as a connective with an > > introduction rule (into something like (G; A |- B). There is such > > a presentation of subtyping with intersection types as the "ISC > > system", a sequent calculus for deriving intersection subtypings, in > > > > Intersection Subtyping with Constructors > > Olivier Laurent, 2018 > > > > > > On Tue, Jan 8, 2019 at 9:55 AM Christopher Jenkins > > wrote: > > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > ] > > > > > > In Curry-style type systems, it is possible for a term to have two > > > different types; in the setting that I am interested in (Cedille) it is > > > possible to define a type of "Cast S T" witnessing that all terms of > type S > > > also have type T, which in turn induces a form of subtyping S <: T. > > > > > > I would like to know whether anyone knows about or can point me to work > > > related to the following two subjects: > > > > > > 1. type systems in which the subtyping judgment is replaced by a type > (like > > > Cast) and where subtyping inference rules elaborate terms with inserted > > > coercions (Luo's "Coercive Subtyping" is closely related in this second > > > respect) > > > > > > 2. the connection between subtyping and positivity checking for > recursive > > > types. To expand on this, let F be some type-scheme of kind Type -> > Type. > > > Often (such as when working with inductive data-types as least > fixpoints of > > > functors) one is interested in restricting F to be positive. With Cast > as > > > the notion of subtyping, this requirement can be stated as the > existence of > > > a proof / function of type X <: Y -> F X <: F Y, in effect allowing > one to > > > replace the usual detailed account of positivity checking with a set of > > > (elaborating) subtyping rules and saying "F is positive iff F X <: F Y > is > > > derivable assuming X <: Y is in the context". > > > > > > Maybe this second observation isn't very deep, but nonetheless I could > not > > > find any work reducing positivity checking to a subtyping problem, and > > > wanted to know if someone else has also made this connection. > > > > > > Thanks, > > > > From cwjnkins at gmail.com Tue Jan 8 15:43:03 2019 From: cwjnkins at gmail.com (Christopher Jenkins) Date: Tue, 8 Jan 2019 14:43:03 -0600 Subject: [TYPES] Literature relating subtyping to positivity checking for recursive types? In-Reply-To: References: Message-ID: > > Maybe silly question: What about F X = X, or beta-equivalent > functions? That is covariant (satisfies `X <: Y -> F X <: F Y`), but > mu F is not well-defined because F is not *contractive*. > Do I miss something, or is that *the* obvious and implied exception? > (Feel free to answer on-list if the question isn't too silly). > > Cheers, It's not a silly question at all! I did some experimentation in Cedille and the answer is that for our derived notion of functor fix-point you can indeed take the fixpoint of `F X = X' -- but the resulting type is uninhabited (you can construct a function of type mu F -> ? X: Type. X.) On Tue, Jan 8, 2019 at 9:16 AM Paolo Giarrusso wrote: > On Tue, 8 Jan 2019 at 09:55, Christopher Jenkins > wrote: > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > In Curry-style type systems, it is possible for a term to have two > > different types; in the setting that I am interested in (Cedille) it is > > possible to define a type of "Cast S T" witnessing that all terms of > type S > > also have type T, which in turn induces a form of subtyping S <: T. > > > 2. the connection between subtyping and positivity checking for recursive > > types. To expand on this, let F be some type-scheme of kind Type -> Type. > > Often (such as when working with inductive data-types as least fixpoints > of > > functors) one is interested in restricting F to be positive. With Cast as > > the notion of subtyping, this requirement can be stated as the existence > of > > a proof / function of type X <: Y -> F X <: F Y, in effect allowing one > to > > replace the usual detailed account of positivity checking with a set of > > (elaborating) subtyping rules and saying "F is positive iff F X <: F Y is > > derivable assuming X <: Y is in the context". > > Maybe silly question: What about F X = X, or beta-equivalent > functions? That is covariant (satisfies `X <: Y -> F X <: F Y`), but > mu F is not well-defined because F is not *contractive*. > Do I miss something, or is that *the* obvious and implied exception? > (Feel free to answer on-list if the question isn't too silly). > > Cheers, > -- > Paolo G. Giarrusso > From anstenklev at gmail.com Tue Feb 5 03:18:30 2019 From: anstenklev at gmail.com (=?UTF-8?Q?Ansten_M=C3=B8rch_Klev?=) Date: Tue, 5 Feb 2019 09:18:30 +0100 Subject: [TYPES] =?utf-8?q?G=C3=B6del=27s_T_extended_with_an_eta-like_rul?= =?utf-8?q?e_for_the_recursor?= Message-ID: Let N be the natural numbers and A any type in the hierarchy of simple types over N. Let s : (N)N be the successor function and let f : (N)A be arbitrary. Then, for variables x : N, y : A we may form the term [x, y]f(s(x)) : (N)(A)A Let R be the recursor for type A. Then, for any term n : N, we have R( f(0) , [x, y]f(s(x)) , n ) : A By the reduction rules for R one can see that f and R( f(0) , [x, y]f(s(x)) ) agree on the numerals. Suppose we add the following eta-like reduction rule to G?del's T: R( f(0) , [x, y]f(s(x)) , x ) --> f(x) Is it known whether the resulting system is (strongly) normalizing and confluent? ------------ Ansten Klev Czech Academy of Sciences, Department of Philosophy From frederic.blanqui at inria.fr Tue Feb 5 04:24:35 2019 From: frederic.blanqui at inria.fr (=?UTF-8?B?RnLDqWTDqXJpYyBCbGFucXVp?=) Date: Tue, 5 Feb 2019 10:24:35 +0100 Subject: [TYPES] =?utf-8?q?G=C3=B6del=27s_T_extended_with_an_eta-like_rul?= =?utf-8?q?e_for_the_recursor?= In-Reply-To: References: Message-ID: <0472c523-33ad-f342-45b5-1185b04d7e90@inria.fr> Hello. Okada and Scott proved that it terminates and is ground confluent but not confluent: https://keio.pure.elsevier.com/en/publications/a-note-on-rewriting-theory-for-uniqueness-of-iteration A note on rewriting theory for uniqueness of iteration Mitsuhiro Okada, P. J. Scott Abstract Uniqueness for higher type term constructors in lambda calculi (e.g. surjective pairing for product types, or uniqueness of iterators on the natural numbers) is easily expressed using universally quantified conditional equations. We use a technique of Lambek [18] involving Mal'cev operators to equationally express uniqueness of iteration (more generally, higher-order primitive recursion) in a simply typed lambda calculus, essentially Godel's T [29,13]. We prove the following facts about typed lambda calculus with uniqueness for primitive recursors: (i) It is undecidable, (ii) Church-Rosser fails, although ground Church-Rosser holds, (iii) strong normalization (termination) is still valid. This entails the undecidability of the coherence problem for cartesian closed categories with strong natural numbers objects, as well as providing a natural example of the following computational paradigm: a non-CR, ground CR, undecidable, terminating rewriting system. Pages ??? 47-64 Number of pages ??? 18 Journal ??? Theory and Applications of Categories Volume ??? 6 Publication status ??? Published - 2000 Le 05/02/2019 ? 09:18, Ansten M?rch Klev a ?crit?: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Let N be the natural numbers and A any type in the hierarchy of simple > types over N. Let s : (N)N be the successor function and let f : (N)A be > arbitrary. Then, for variables x : N, y : A we may form the term > > [x, y]f(s(x)) : (N)(A)A > > Let R be the recursor for type A. Then, for any term n : N, we have > > R( f(0) , [x, y]f(s(x)) , n ) : A > > By the reduction rules for R one can see that f and R( f(0) , [x, y]f(s(x)) > ) agree on the numerals. > > Suppose we add the following eta-like reduction rule to G?del's T: > > R( f(0) , [x, y]f(s(x)) , x ) --> f(x) > > Is it known whether the resulting system is (strongly) normalizing and > confluent? > > > ------------ > Ansten Klev > Czech Academy of Sciences, Department of Philosophy From danel.ahman at eesti.ee Tue Feb 5 05:21:10 2019 From: danel.ahman at eesti.ee (Danel Ahman) Date: Tue, 5 Feb 2019 11:21:10 +0100 Subject: [TYPES] =?utf-8?q?G=C3=B6del=27s_T_extended_with_an_eta-like_rul?= =?utf-8?q?e_for_the_recursor?= In-Reply-To: References: Message-ID: <5E745A93-3CFF-4177-9564-035459B0E5D5@eesti.ee> Dear Ansten, I believe your questions are answered in this paper from Okada and Scott http://www.tac.mta.ca/tac/volumes/6/n4/6-04abs.html Best, Danel > On 5 Feb 2019, at 09:18, Ansten M?rch Klev wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Let N be the natural numbers and A any type in the hierarchy of simple > types over N. Let s : (N)N be the successor function and let f : (N)A be > arbitrary. Then, for variables x : N, y : A we may form the term > > [x, y]f(s(x)) : (N)(A)A > > Let R be the recursor for type A. Then, for any term n : N, we have > > R( f(0) , [x, y]f(s(x)) , n ) : A > > By the reduction rules for R one can see that f and R( f(0) , [x, y]f(s(x)) > ) agree on the numerals. > > Suppose we add the following eta-like reduction rule to G?del's T: > > R( f(0) , [x, y]f(s(x)) , x ) --> f(x) > > Is it known whether the resulting system is (strongly) normalizing and > confluent? > > > ------------ > Ansten Klev > Czech Academy of Sciences, Department of Philosophy From Sergei.Soloviev at irit.fr Tue Feb 5 05:28:30 2019 From: Sergei.Soloviev at irit.fr (Sergei Soloviev) Date: Tue, 05 Feb 2019 11:28:30 +0100 Subject: [TYPES] =?utf-8?b?Pz09P3V0Zi04P3E/ID89PT91dGYtOD9xPyBHw7ZkZWwn?= =?utf-8?q?s_T_extended_with_an_eta-like_rule_for_the_recursor?= In-Reply-To: Message-ID: Dear Ansten, We did work with David Chemouil on related questions, main results and ideas are in the papers David Chemouil: Isomorphisms of simple inductive types through extensional rewriting. Mathematical Structures in Computer Science 15(5): 875-915 (2005) David Chemouil: An insertion operator preserving infinite reduction sequences. Mathematical Structures in Computer Science 18(4): 693-728 (2008) David Chemouil, Sergei Soloviev: Remarks on isomorphisms of simple inductive types. Electr. Notes Theor. Comput. Sci. 85(7): 106-124 (2003) I have to check whether they cover directly your question, but in any case there were developped techniques that certainly could be useful. All the best Sergei Soloviev Le Mardi 5 F?vrier 2019 09:18 CET, Ansten M?rch Klev a ?crit: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Let N be the natural numbers and A any type in the hierarchy of simple > types over N. Let s : (N)N be the successor function and let f : (N)A be > arbitrary. Then, for variables x : N, y : A we may form the term > > [x, y]f(s(x)) : (N)(A)A > > Let R be the recursor for type A. Then, for any term n : N, we have > > R( f(0) , [x, y]f(s(x)) , n ) : A > > By the reduction rules for R one can see that f and R( f(0) , [x, y]f(s(x)) > ) agree on the numerals. > > Suppose we add the following eta-like reduction rule to G?del's T: > > R( f(0) , [x, y]f(s(x)) , x ) --> f(x) > > Is it known whether the resulting system is (strongly) normalizing and > confluent? > > > ------------ > Ansten Klev > Czech Academy of Sciences, Department of Philosophy From David.Chemouil at onera.fr Tue Feb 5 06:42:56 2019 From: David.Chemouil at onera.fr (David CHEMOUIL) Date: Tue, 05 Feb 2019 12:42:56 +0100 Subject: [TYPES] =?utf-8?q?G=C3=B6del=27s_T_extended_with_an_eta-like_rul?= =?utf-8?q?e_for_the_recursor?= In-Reply-To: References: Message-ID: <5e6b9e7532d3fd3dc493bbfcf350dc50@sybille.onecert.fr> Dear Ansten, [Okada and Scott 1999] is certainly the paper to read here. The abstract goes: " Uniqueness for higher type term constructors in lambda calculi (e.g. surjective pairing for product types, or uniqueness of iterators on the natural numbers) is easily expressed using universally quantified conditional equations. We use a technique of Lambek [18] involving Mal'cev operators to equationally express uniqueness of iteration (more generally, higher-order primitive recursion) in a simply typed lambda calculus, essentially Godel's T [29,13]. We prove the following facts about typed lambda calculus with uniqueness for primitive recursors: (i) It is undecidable, (ii) Church-Rosser fails, although ground Church-Rosser holds, (iii) strong normalization (termination) is still valid. This entails the undecidability of the coherence problem for cartesian closed categories with strong natural numbers objects, as well as providing a natural example of the following computational paradigm: a non-CR, ground CR, undecidable, terminating rewriting system. " [Okada and Scott 1999] A Note on Rewriting Theory for Uniqueness of Iteration http://www.tac.mta.ca/tac/volumes/6/n4/6-04abs.html Best regards, david Le 2019-02-05 11:28, Sergei Soloviev a ?crit?: > Dear Ansten, > > We did work with David Chemouil on related questions, main results and > ideas > are in the papers > > David Chemouil: > Isomorphisms of simple inductive types through extensional rewriting. > Mathematical Structures in Computer Science 15(5): 875-915 (2005) > > David Chemouil: > An insertion operator preserving infinite reduction sequences. > Mathematical Structures in Computer Science 18(4): 693-728 (2008) > > David Chemouil, Sergei Soloviev: > Remarks on isomorphisms of simple inductive types. Electr. Notes > Theor. Comput. Sci. 85(7): 106-124 (2003) > > I have to check whether they cover directly your question, but in any > case there were > developped techniques that certainly could be useful. > > All the best > > Sergei Soloviev > > > Le Mardi 5 F?vrier 2019 09:18 CET, Ansten M?rch Klev > a ?crit: > >> [ The Types Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> >> Let N be the natural numbers and A any type in the hierarchy of simple >> types over N. Let s : (N)N be the successor function and let f : (N)A >> be >> arbitrary. Then, for variables x : N, y : A we may form the term >> >> [x, y]f(s(x)) : (N)(A)A >> >> Let R be the recursor for type A. Then, for any term n : N, we have >> >> R( f(0) , [x, y]f(s(x)) , n ) : A >> >> By the reduction rules for R one can see that f and R( f(0) , [x, >> y]f(s(x)) >> ) agree on the numerals. >> >> Suppose we add the following eta-like reduction rule to G?del's T: >> >> R( f(0) , [x, y]f(s(x)) , x ) --> f(x) >> >> Is it known whether the resulting system is (strongly) normalizing and >> confluent? >> >> >> ------------ >> Ansten Klev >> Czech Academy of Sciences, Department of Philosophy -- David Chemouil ONERA DTIS & Universit? de Toulouse tel:+33-5-6225-2936 From monnier at iro.umontreal.ca Tue Feb 5 17:47:57 2019 From: monnier at iro.umontreal.ca (Stefan Monnier) Date: Tue, 05 Feb 2019 17:47:57 -0500 Subject: [TYPES] Conversion irrelevance and extensionality Message-ID: In the context of ICC/ICC* where conversion is strengthened by ignoring erasable arguments, I bumped into the following comment: as soon as one considers the extension to inductive types where conversion irrelevance provides some form of weak extensionality. Does someone here know what the above might be referring to? Stefan From dankoilik at gmail.com Sun Feb 10 12:47:46 2019 From: dankoilik at gmail.com (Danko ILIK) Date: Sun, 10 Feb 2019 18:47:46 +0100 Subject: [TYPES] =?iso-8859-1?q?G=F6del=27s_T_extended_with_an_eta-like_r?= =?iso-8859-1?q?ule_for_the_recursor?= In-Reply-To: References: Message-ID: <20190210174746.GA6850@thinkc.home> On Tue, Feb 05, 2019 at 09:18:30AM +0100, Ansten M?rch Klev wrote: > Suppose we add the following eta-like reduction rule to G?del's T: > > R( f(0) , [x, y]f(s(x)) , x ) --> f(x) > > Is it known whether the resulting system is (strongly) normalizing and > confluent? Dear Ansten, Maybe adding this kind of eta rule is not problematic when A is to vary over simple types only. I don't know. However, in "reality" (in proof assistants, in programming languages, or even category theory) we often have to consider richer type structures, such as inductive types. And, in this case it still seems problematic to add eta rules, even for special cases of the recursor which is the case distinction operator for co-product types. If you want to browse through that literature, I can point to the Related Works section of my paper https://arxiv.org/abs/1502.04634 which should mention most of the works on that problem. Best regards, Danko From brianberns at gmail.com Wed Mar 27 21:49:56 2019 From: brianberns at gmail.com (Brian Berns) Date: Wed, 27 Mar 2019 21:49:56 -0400 Subject: [TYPES] Order of evaluation rules in untyped lambda-calculus Message-ID: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> I'm working through Pierce's _Types and Programming Languages_ and I've found a subtle issue that I'd could use some help on. The problem is with the untyped lambda-calculus. The E-App1 evaluation rule on p. 72 says that t1 t2 -> t1' t2 if t1 -> t1' with the following comment: "Notice how the choice of metavariables in these rules helps control the order of evaluation. ... Similarly, rule E-App1 applies to any application whose left-hand side is not a value, since t1 can match any term whatsoever, but **the premise further requires that t1 can take a step**." (Emphasis added.) This strongly implies that the order of the rules shouldn't matter. The corresponding implementation on p. 87 then says "The single-step evaluation function is a direct transcription of the evaluation rules", but the rules appear in a different order and there is no guard on the E-App1 rule that prevents it from firing when t1 can't be reduced. Instead, it looks like the rules are arranged in an order that ensures that E-App1 is executed only as a last resort. It seems to me that the "correct" implementation of E-App1 (and, in fact, of every evaluation rule) is to ensure that its premises are met before applying it. Instead, the implementation seems to take a shortcut here. I'm not opposed to that, but I'd like to understand how it works. Am I correct in thinking that the behavior of this implementation is subtly dependent on the order of its evaluation rules in a way that the definition of those rules was intended to avoid? If that's the case, are there any general guidelines that an implementor can/should use to order the evaluation rules for a language in the correct way? Thanks for your help. -- Brian Berns From clements at brinckerhoff.org Sun Mar 31 12:31:40 2019 From: clements at brinckerhoff.org (John Clements) Date: 31 Mar 2019 12:31:40 -0400 Subject: [TYPES] Order of evaluation rules in untyped lambda-calculus In-Reply-To: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> References: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> Message-ID: <19b7ff0b-1563-45d6-9bdc-34767a1f1d56@mtasv.net> Forgive me for answering your question with a question: Is there a difference between the two systems? That is, does the implementation relate two terms that the formal system does not, or fail to relate two terms that the formal system does? John Clements > On Mar 27, 2019, at 18:49, Brian Berns wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > I'm working through Pierce's _Types and Programming Languages_ and I've > found a subtle issue that I'd could use some help on. The problem is with > the untyped lambda-calculus. The E-App1 evaluation rule on p. 72 says that > t1 t2 -> t1' t2 if t1 -> t1' with the following comment: > > "Notice how the choice of metavariables in these rules helps control the > order of evaluation. ... Similarly, rule E-App1 applies to any application > whose left-hand side is not a value, since t1 can match any term whatsoever, > but **the premise further requires that t1 can take a step**." (Emphasis > added.) > > This strongly implies that the order of the rules shouldn't matter. The > corresponding implementation on p. 87 then says "The single-step evaluation > function is a direct transcription of the evaluation rules", but the rules > appear in a different order and there is no guard on the E-App1 rule that > prevents it from firing when t1 can't be reduced. Instead, it looks like the > rules are arranged in an order that ensures that E-App1 is executed only as > a last resort. > > It seems to me that the "correct" implementation of E-App1 (and, in fact, of > every evaluation rule) is to ensure that its premises are met before > applying it. Instead, the implementation seems to take a shortcut here. I'm > not opposed to that, but I'd like to understand how it works. Am I correct > in thinking that the behavior of this implementation is subtly dependent on > the order of its evaluation rules in a way that the definition of those > rules was intended to avoid? If that's the case, are there any general > guidelines that an implementor can/should use to order the evaluation rules > for a language in the correct way? > > Thanks for your help. > > -- Brian Berns > From brianberns at gmail.com Sun Mar 31 12:43:54 2019 From: brianberns at gmail.com (Brian Berns) Date: Sun, 31 Mar 2019 12:43:54 -0400 Subject: [TYPES] Order of evaluation rules in untyped lambda-calculus In-Reply-To: <19b7ff0b-1563-45d6-9bdc-34767a1f1d56@mtasv.net> References: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> <19b7ff0b-1563-45d6-9bdc-34767a1f1d56@mtasv.net> Message-ID: <005e01d4e7e0$edc4df60$c94e9e20$@gmail.com> I think his implementation produces the same results as the formal system, although I haven't verified this. It just does it in a way that is both different from what he explicitly describes, and significantly harder to reason about (IMHO). -- Brian -----Original Message----- From: John Clements Sent: Sunday, March 31, 2019 12:32 PM To: Brian Berns Cc: Subject: Re: [TYPES] Order of evaluation rules in untyped lambda-calculus Forgive me for answering your question with a question: Is there a difference between the two systems? That is, does the implementation relate two terms that the formal system does not, or fail to relate two terms that the formal system does? John Clements > On Mar 27, 2019, at 18:49, Brian Berns wrote: > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > I'm working through Pierce's _Types and Programming Languages_ and > I've found a subtle issue that I'd could use some help on. The problem > is with the untyped lambda-calculus. The E-App1 evaluation rule on p. > 72 says that > t1 t2 -> t1' t2 if t1 -> t1' with the following comment: > > "Notice how the choice of metavariables in these rules helps control > the order of evaluation. ... Similarly, rule E-App1 applies to any > application whose left-hand side is not a value, since t1 can match > any term whatsoever, but **the premise further requires that t1 can > take a step**." (Emphasis > added.) > > This strongly implies that the order of the rules shouldn't matter. > The corresponding implementation on p. 87 then says "The single-step > evaluation function is a direct transcription of the evaluation > rules", but the rules appear in a different order and there is no > guard on the E-App1 rule that prevents it from firing when t1 can't be > reduced. Instead, it looks like the rules are arranged in an order > that ensures that E-App1 is executed only as a last resort. > > It seems to me that the "correct" implementation of E-App1 (and, in > fact, of every evaluation rule) is to ensure that its premises are met > before applying it. Instead, the implementation seems to take a > shortcut here. I'm not opposed to that, but I'd like to understand how > it works. Am I correct in thinking that the behavior of this > implementation is subtly dependent on the order of its evaluation > rules in a way that the definition of those rules was intended to > avoid? If that's the case, are there any general guidelines that an > implementor can/should use to order the evaluation rules for a language in the correct way? > > Thanks for your help. > > -- Brian Berns > From clements at brinckerhoff.org Sun Mar 31 12:55:50 2019 From: clements at brinckerhoff.org (John Clements) Date: 31 Mar 2019 12:55:50 -0400 Subject: [TYPES] Order of evaluation rules in untyped lambda-calculus In-Reply-To: <005e01d4e7e0$edc4df60$c94e9e20$@gmail.com> References: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> <19b7ff0b-1563-45d6-9bdc-34767a1f1d56@mtasv.net> <005e01d4e7e0$edc4df60$c94e9e20$@gmail.com> Message-ID: Is the system you?re thinking about one whose conditionals would attempt to verify premises before deciding which rule to use? If so, I would point out that the system as written makes different things easier to reason about: specifically, the given implementation clearly does not explore multiple different possibilities, and therefore defines evaluation to be a function (at most one output for a given input). Also, it gives a much clearer sense of what the computational cost of generating the answer is going to be. Naturally, you?re welcome to disagree :). John > On Mar 31, 2019, at 09:43, Brian Berns wrote: > > I think his implementation produces the same results as the formal system, although I haven't verified this. It just does it in a way that is both different from what he explicitly describes, and significantly harder to reason about (IMHO). > > -- Brian > > -----Original Message----- > From: John Clements > Sent: Sunday, March 31, 2019 12:32 PM > To: Brian Berns > Cc: > Subject: Re: [TYPES] Order of evaluation rules in untyped lambda-calculus > > Forgive me for answering your question with a question: Is there a difference between the two systems? That is, does the implementation relate two terms that the formal system does not, or fail to relate two terms that the formal system does? > > John Clements > >> On Mar 27, 2019, at 18:49, Brian Berns wrote: >> >> [ The Types Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> >> I'm working through Pierce's _Types and Programming Languages_ and >> I've found a subtle issue that I'd could use some help on. The problem >> is with the untyped lambda-calculus. The E-App1 evaluation rule on p. >> 72 says that >> t1 t2 -> t1' t2 if t1 -> t1' with the following comment: >> >> "Notice how the choice of metavariables in these rules helps control >> the order of evaluation. ... Similarly, rule E-App1 applies to any >> application whose left-hand side is not a value, since t1 can match >> any term whatsoever, but **the premise further requires that t1 can >> take a step**." (Emphasis >> added.) >> >> This strongly implies that the order of the rules shouldn't matter. >> The corresponding implementation on p. 87 then says "The single-step >> evaluation function is a direct transcription of the evaluation >> rules", but the rules appear in a different order and there is no >> guard on the E-App1 rule that prevents it from firing when t1 can't be >> reduced. Instead, it looks like the rules are arranged in an order >> that ensures that E-App1 is executed only as a last resort. >> >> It seems to me that the "correct" implementation of E-App1 (and, in >> fact, of every evaluation rule) is to ensure that its premises are met >> before applying it. Instead, the implementation seems to take a >> shortcut here. I'm not opposed to that, but I'd like to understand how >> it works. Am I correct in thinking that the behavior of this >> implementation is subtly dependent on the order of its evaluation >> rules in a way that the definition of those rules was intended to >> avoid? If that's the case, are there any general guidelines that an >> implementor can/should use to order the evaluation rules for a language in the correct way? >> >> Thanks for your help. >> >> -- Brian Berns >> > > > > From ezra at ezrakilty.net Sun Mar 31 13:04:52 2019 From: ezra at ezrakilty.net (Ezra e. k. Cooper) Date: Sun, 31 Mar 2019 13:04:52 -0400 Subject: [TYPES] Order of evaluation rules in untyped lambda-calculus In-Reply-To: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> References: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> Message-ID: <27e45d75-fed7-48db-8705-2755d3d4162e@www.fastmail.com> It's a good question! Something subtle is indeed going on. (Or more than one.) In most such systems, the value grammar is carefully chosen so that it does not reduce, and that all "meaningful" terms which do not reduce are in this grammar. Type systems will often play a role in ensuring that well-typed terms either reduce or are values. If you know that the values do not reduce and that the meaningful values do, then you know that the terms t1 for which t1 -> t1' are exactly the non-values (presuming the whole term is meaningful). It would usually be a bug in a semantics to define the values so that any of them reduce, but it's a proof obligation to show that you don't have that bug. Strictly speaking, you are right that a little shortcut has been taken in the implementation. Strictly, to determine if an application admits the reduction step of E-App1, we should somehow check that the LHS reduces. How can we do that within this implementation? Well it happens that all the cases produce a new term (representing a reduction) except that final case whose body reads "raise NoRuleApplies". So, if we were auto-generating this sort of evaluator from the inference rules, we might implement the case for E-App1 by *trying* to evaluate the LHS and catching that exception. If there is no exception then we get the new left-hand term, t1'. If there is an exception then the rule E-App1 does not apply and we should fall through to other cases. Implementing all that as a pattern guard in OCaml would be a bit fiddly, though. If you imagine implementing this system in, say, Prolog, you would very naturally write it so that some reduction exists whenever some other reduction exists; the interpreter knows how to search for those reductions, and exactly what to do if it doesn't find one (i.e., try to meet its goal through other rules, which is the equivalent of "falling through" in this case). That would be a more "direct" implementation of the rewrite rules, perhaps. The activity that the Pierce evaluator is doing here is called "redex selection," or focusing, and, in a sense the evaluator on p. 87 is already a little "virtual machine" that has smarts about redex selection. There is a significant literature on virtual machines and how they correspond to rewrite rules as such. Danvy and Nielsen's "Refocusing in reduction semantics" is one notable paper in that area. Cheers, Ezra On Sun, Mar 31, 2019, at 7:13 AM, Brian Berns wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > I'm working through Pierce's _Types and Programming Languages_ and I've > found a subtle issue that I'd could use some help on. The problem is with > the untyped lambda-calculus. The E-App1 evaluation rule on p. 72 says that > t1 t2 -> t1' t2 if t1 -> t1' with the following comment: > > "Notice how the choice of metavariables in these rules helps control the > order of evaluation. ... Similarly, rule E-App1 applies to any application > whose left-hand side is not a value, since t1 can match any term whatsoever, > but **the premise further requires that t1 can take a step**." (Emphasis > added.) > > This strongly implies that the order of the rules shouldn't matter. The > corresponding implementation on p. 87 then says "The single-step evaluation > function is a direct transcription of the evaluation rules", but the rules > appear in a different order and there is no guard on the E-App1 rule that > prevents it from firing when t1 can't be reduced. Instead, it looks like the > rules are arranged in an order that ensures that E-App1 is executed only as > a last resort. > > It seems to me that the "correct" implementation of E-App1 (and, in fact, of > every evaluation rule) is to ensure that its premises are met before > applying it. Instead, the implementation seems to take a shortcut here. I'm > not opposed to that, but I'd like to understand how it works. Am I correct > in thinking that the behavior of this implementation is subtly dependent on > the order of its evaluation rules in a way that the definition of those > rules was intended to avoid? If that's the case, are there any general > guidelines that an implementor can/should use to order the evaluation rules > for a language in the correct way? > > Thanks for your help. > > -- Brian Berns > > From brianberns at gmail.com Sun Mar 31 13:36:44 2019 From: brianberns at gmail.com (Brian Berns) Date: Sun, 31 Mar 2019 13:36:44 -0400 Subject: [TYPES] Order of evaluation rules in untyped lambda-calculus In-Reply-To: References: <027b01d4e508$8c1fa450$a45eecf0$@gmail.com> <19b7ff0b-1563-45d6-9bdc-34767a1f1d56@mtasv.net> <005e01d4e7e0$edc4df60$c94e9e20$@gmail.com> Message-ID: <008101d4e7e8$4f19b220$ed4d1660$@gmail.com> Yes, I'm verifying premises before deciding which rule to use. I do find that easier to think about, and potentially less buggy, although I totally agree that it could also be slower in practice. (I'm only building a toy implementation to crystalize my understanding of the material, so performance isn't critical in my case.) I believe that my evaluation routine should still be a valid function, because the formal system defines the rules so that no more than one applies to any given term. FWIW, I'm implementing this system in F#, which has a feature called "active patterns" that supports a near-literal implementation of the spec. It looks something like this: /// Active pattern for a term that can take a step. let rec (|Step|_|) term = step term /// Reduces a term by one step, if it's not already in normal form. and step = function // try to reduce first term | Application (Step t1', t2) -> // <== key line is here Application (t1', t2) |> Some The key thing here is that the Application pattern first checks that t1 reduces to t1' before matching. Thanks for your feedback. I appreciate it. -- Brian -----Original Message----- From: John Clements Sent: Sunday, March 31, 2019 12:56 PM To: Brian Berns Cc: Subject: Re: [TYPES] Order of evaluation rules in untyped lambda-calculus Is the system you?re thinking about one whose conditionals would attempt to verify premises before deciding which rule to use? If so, I would point out that the system as written makes different things easier to reason about: specifically, the given implementation clearly does not explore multiple different possibilities, and therefore defines evaluation to be a function (at most one output for a given input). Also, it gives a much clearer sense of what the computational cost of generating the answer is going to be. Naturally, you?re welcome to disagree :). John > On Mar 31, 2019, at 09:43, Brian Berns wrote: > > I think his implementation produces the same results as the formal system, although I haven't verified this. It just does it in a way that is both different from what he explicitly describes, and significantly harder to reason about (IMHO). > > -- Brian > > -----Original Message----- > From: John Clements > Sent: Sunday, March 31, 2019 12:32 PM > To: Brian Berns > Cc: > > Subject: Re: [TYPES] Order of evaluation rules in untyped > lambda-calculus > > Forgive me for answering your question with a question: Is there a difference between the two systems? That is, does the implementation relate two terms that the formal system does not, or fail to relate two terms that the formal system does? > > John Clements > >> On Mar 27, 2019, at 18:49, Brian Berns wrote: >> >> [ The Types Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> >> I'm working through Pierce's _Types and Programming Languages_ and >> I've found a subtle issue that I'd could use some help on. The >> problem is with the untyped lambda-calculus. The E-App1 evaluation rule on p. >> 72 says that >> t1 t2 -> t1' t2 if t1 -> t1' with the following comment: >> >> "Notice how the choice of metavariables in these rules helps control >> the order of evaluation. ... Similarly, rule E-App1 applies to any >> application whose left-hand side is not a value, since t1 can match >> any term whatsoever, but **the premise further requires that t1 can >> take a step**." (Emphasis >> added.) >> >> This strongly implies that the order of the rules shouldn't matter. >> The corresponding implementation on p. 87 then says "The single-step >> evaluation function is a direct transcription of the evaluation >> rules", but the rules appear in a different order and there is no >> guard on the E-App1 rule that prevents it from firing when t1 can't >> be reduced. Instead, it looks like the rules are arranged in an order >> that ensures that E-App1 is executed only as a last resort. >> >> It seems to me that the "correct" implementation of E-App1 (and, in >> fact, of every evaluation rule) is to ensure that its premises are >> met before applying it. Instead, the implementation seems to take a >> shortcut here. I'm not opposed to that, but I'd like to understand >> how it works. Am I correct in thinking that the behavior of this >> implementation is subtly dependent on the order of its evaluation >> rules in a way that the definition of those rules was intended to >> avoid? If that's the case, are there any general guidelines that an >> implementor can/should use to order the evaluation rules for a language in the correct way? >> >> Thanks for your help. >> >> -- Brian Berns >> > > > > From beta.ziliani at gmail.com Thu Apr 11 11:08:22 2019 From: beta.ziliani at gmail.com (Beta Ziliani) Date: Thu, 11 Apr 2019 12:08:22 -0300 Subject: [TYPES] =?utf-8?q?How_to_code_interesting_examples_in_=CE=BBC?= Message-ID: Hi, In a course I'm teaching, I wanted to show the Calculus of Constructions (a.k.a. ?C) prior to jumping to the Calculus of Inductive Constructions (CIC). However, I find myself unable to code any interesting example. In particular, I wasn't able to Church-encode a vector, a list with *n* elements, whose *n* is defined in the type. Or, more primitively, the equality type. Someone knows if this is possible to do, or send me pointers at where such problem is discussed? Thanks, Beta From herman at cs.ru.nl Thu Apr 11 11:24:41 2019 From: herman at cs.ru.nl (Herman Geuvers) Date: Thu, 11 Apr 2019 17:24:41 +0200 Subject: [TYPES] =?utf-8?q?How_to_code_interesting_examples_in_=CE=BBC?= In-Reply-To: References: Message-ID: <40b274cf-cd2b-8cba-5503-5af0491fb754@cs.ru.nl> Hi Beta, For my proving with Computer Assistance course I let the students do higher order logic exercises in the Calculus of Constructions. Define Leibniz equality and prove properties, define the transitive closure and prove properties etc. See http://www.cs.ru.nl/~herman/onderwijs/provingwithCA/exercises10.pdf Best Herman On 4/11/19 5:08 PM, Beta Ziliani wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Hi, > > In a course I'm teaching, I wanted to show the Calculus of Constructions > (a.k.a. ?C) prior to jumping to the Calculus of Inductive Constructions > (CIC). However, I find myself unable to code any interesting example. In > particular, I wasn't able to Church-encode a vector, a list with *n* > elements, whose *n* is defined in the type. Or, more primitively, the > equality type. Someone knows if this is possible to do, or send me pointers > at where such problem is discussed? > > Thanks, > Beta > From beta.ziliani at gmail.com Mon Apr 15 11:17:45 2019 From: beta.ziliani at gmail.com (Beta Ziliani) Date: Mon, 15 Apr 2019 12:17:45 -0300 Subject: [TYPES] =?utf-8?q?How_to_code_interesting_examples_in_=CE=BBC?= In-Reply-To: References: Message-ID: Thanks Herman and Randy! Very nice pointers to start with. I am still failing to see how to properly encode the type of a vector, without extrinsically defining a list with a proof of its size. But I admit I haven't tried hard. In any case, your pointers suffice for my purpose now. On Thu, Apr 11, 2019 at 6:42 PM R. Pollack wrote: > Go to the source: > > Coquand and Huet, "Constructions: A Higher Order Proof System for > Mechanizing Mathematics". > Eurocal '85, LNCS 203 > > --Randy > > On Thu, Apr 11, 2019 at 4:28 PM Beta Ziliani > wrote: > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > Hi, > > > > In a course I'm teaching, I wanted to show the Calculus of Constructions > > (a.k.a. ?C) prior to jumping to the Calculus of Inductive Constructions > > (CIC). However, I find myself unable to code any interesting example. In > > particular, I wasn't able to Church-encode a vector, a list with *n* > > elements, whose *n* is defined in the type. Or, more primitively, the > > equality type. Someone knows if this is possible to do, or send me > pointers > > at where such problem is discussed? > > > > Thanks, > > Beta > From xhli06 at 163.com Sun May 26 21:27:26 2019 From: xhli06 at 163.com (Xuhui Li) Date: Mon, 27 May 2019 09:27:26 +0800 (CST) Subject: [TYPES] Why cannot declare a super type dynamically? Message-ID: <2d2dff34.9902.16af6e6423d.Coremail.xhli06@163.com> Dear All, As far as I know, the existing type systems allow the subtyping being declared in the defintion of the sub types but forbid the super type being declared after the subtype . For example, class A extends B. However, if a new type C is declared and it is found that conceputually C should be a supertype of A (assume that the fields of C don't occur in A but an injection of the fields can be defined) , we cannot declare A extends C after A has been already defined. Why cannot this happen? Will dynamically introducing a supertype threaten the safety of existing type system? Thanks for your attention. Best Regards Xuhui From jeremy.siek at gmail.com Mon May 27 12:34:52 2019 From: jeremy.siek at gmail.com (Jeremy Siek) Date: Mon, 27 May 2019 12:34:52 -0400 Subject: [TYPES] Why cannot declare a super type dynamically? In-Reply-To: <2d2dff34.9902.16af6e6423d.Coremail.xhli06@163.com> References: <2d2dff34.9902.16af6e6423d.Coremail.xhli06@163.com> Message-ID: Dear Xuhui, Here?s a few related papers from the late 1990?s. Signatures: A language extension for improving type abstraction and subtype polymorphism in C++ Gerald Baumgartner, Vincent F. Russo https://doi.org/10.1002/spe.4380250803 Implementing signatures for C++. Gerald Baumgartner and Vincent F. Russo. 1997. ACM Trans. Program. Lang. Syst. 19, 1 (January 1997), 153-187. http://dx.doi.org/10.1145/239912.239922 Safe structural conformance for Java. by Laufer, Baumgartner, and Russo. The Computer Journal, 43(6) 2000. Cheers, Jeremy > On May 26, 2019, at 9:27 PM, Xuhui Li wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear All, > > > As far as I know, the existing type systems allow the subtyping being declared in the defintion of the sub types but forbid the super type being declared after the subtype . For example, class A extends B. However, if a new type C is declared and it is found that conceputually C should be a supertype of A (assume that the fields of C don't occur in A but an injection of the fields can be defined) , we cannot declare A extends C after A has been already defined. Why cannot this happen? Will dynamically introducing a supertype threaten the safety of existing type system? > > > Thanks for your attention. > > > Best Regards > > > Xuhui From marco.servetto at gmail.com Mon May 27 17:47:08 2019 From: marco.servetto at gmail.com (Marco Servetto) Date: Tue, 28 May 2019 09:47:08 +1200 Subject: [TYPES] Why cannot declare a super type dynamically? In-Reply-To: <2d2dff34.9902.16af6e6423d.Coremail.xhli06@163.com> References: <2d2dff34.9902.16af6e6423d.Coremail.xhli06@163.com> Message-ID: On a more technical/programming note, focusing on just Java interfaces: not only could be theoretically safe to add interfaces later one, someone told me a trick to actually force plain java to do it: (however, I have never used this trick, if someone have practical experience, please share!) Java classes ad interfaces are loaded when they are used, so is possible to write programs that generate bytecode for interface I while running and before interface I is loaded. So... when interfaces get loaded? not at the same time superclasses get loaded, but much later if even. (Java8 may have changed this somehow) In particular they are loaded if needed for an instanceof test Thus to have a class that have expandable supertypes you can do the following: interface ExpandableA0{} class A implements ExpandableA0{ .... } Then, you can use A freely, just never use ExpandableA0, not even as a type annotation. Then, when you dynamicaly discover you want A to be subtype of interface Foo, you do the following Late.implement(A.class,Foo.class) internally this method will do something on the line of the following pseudocode: implement(Class c, Class i){ String expander="Expandable"+c.getName();//will be ExpandableA int num=classLoader.discoverSmalledUnloadedExpander(expander); classLoader.load(expander+num,"interface "+(expander+num)+" extends "+(expander+(num+1))+", "+i.getName()+"{}") //at the first round it will be "interface ExpandableA0 extends ExpandableA1, Foo {}" } Now, you can do A myA=new A(); Late.implement(A.class,Foo.class) Foo f=(Foo)myA; And the dynamic cast should be a success. This also teach us abut formal soundness: OO type systems try to be modular, this means that even if we do not know about all classes and interfaces in the system, we can say that a part is correct in "isolation" It turns out interfaces do not always need to be considered to see if a part implementing them is correct, if the corrisponding subtyping assumptions are not used. Thus, we could say that our part is correct parametric of what those interfaces can be, and this allows us to dynamically assume subtyping I may be wrong in some part of my reasoning/code, please fell free to prove me wrong, it looks "to easy to be true" On Mon, 27 May 2019 at 18:44, Xuhui Li wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear All, > > > As far as I know, the existing type systems allow the subtyping being declared in the defintion of the sub types but forbid the super type being declared after the subtype . For example, class A extends B. However, if a new type C is declared and it is found that conceputually C should be a supertype of A (assume that the fields of C don't occur in A but an injection of the fields can be defined) , we cannot declare A extends C after A has been already defined. Why cannot this happen? Will dynamically introducing a supertype threaten the safety of existing type system? > > > Thanks for your attention. > > > Best Regards > > > Xuhui From p.ohearn at ucl.ac.uk Fri Jun 14 11:23:24 2019 From: p.ohearn at ucl.ac.uk (O'Hearn, Peter) Date: Fri, 14 Jun 2019 15:23:24 +0000 Subject: [TYPES] Variants and [Park or Scott] fixpoint Induction Message-ID: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> Two methods of reasoning about loops are provided by variants and by (Park or Scott) fixpoint induction. Is there a known relation or non-relation between them? My intuition is that fixpoint induction is not suitable for termination or liveness properties, but I am unsure whether this intuition is correct. The Hoare rule for total correctness of while loops using variants is well explained in the wikipedia article: https://en.wikipedia.org/wiki/Hoare_logic#While_rule_for_total_correctness There, you make sure a quantity in a well-founded set decreases on each loop iteration. Here is Park induction: lfp(F) <= S iff Exists I. FI <= I & I <= S If you think of S as ?spec? and I as ?invariant?, then this can form the basis for reasoning about safety properties (as explained by Cousot here) http://web.mit.edu/16.399/www/lecture_11-b-fixpoints1/Cousot_MIT_2005_Course_11b_4-1.pdf I am a bit worried that my intuition "fixpoint induction is not good for termination? might have some holes in it. In particular, Park induction is used in a known complete proof theory for modal mu-calculus https://eprints.illc.uva.nl/569/1/PP-2016-33.text.pdf and that logic is notable for being able to express liveness properties. I asked a few experts who did not know a way to answer my question above, which is why I am posting it more widely here. In particular, if there is an explanation of how/why fixpoint induciton could be good for reasoning about (say) liveness or termination properties of while loops, I?f be glad to hear about it. Thanks! Peter O'Hearn From julbinb at gmail.com Wed Jun 12 15:46:01 2019 From: julbinb at gmail.com (Julia Belyakova) Date: Wed, 12 Jun 2019 15:46:01 -0400 Subject: [TYPES] What exactly should we call syntax-directed inference rules? Message-ID: Dear All, It seems that there is a slight inconsistency in what kind of inference rules are called "syntax-directed". Let's take subtyping rules as an example. (a) In TAPL, syntax-directed means that for each pair of types T and S, there is exactly one rule with the conclusion that matches T <: S. (b) In some papers, e.g. [1], rules are considered syntax-directed if for each pair T and S there is a finite number of rules matching T <: S and each of them has a finite number of (computable) premises. The transitivity rule T1 <: T2 T2 <: T3 ---------------------- T1 <: T3 is clearly bad, and it is rejected by both definitions: (a) The conclusion T1 <: T3 overlaps with any other subtyping rule. (b) It is not clear how to compute T2 and there can be infinitely many of them. But what about subtyping of union types? T <: S1 T <: S2 ------------- UnionR1 ------------ UnionR2 T <: S1?S2 T <: S1?S2 According to (a), such rules are not syntax-directed. According to (b), they are. The UnionR* rules are definitely worse than syntax-directed rules in the sense of (a) because for a triple of types T, S1, S2 we have two options, and one of them might work while the other does not. Algorithmically, this is the source of backtracking. However, these rules are better than the transitivity rule because their premises are uniquely defined by syntax. Which terminology is more generally accepted, (a) or (b)? And is there a suitable terminology to distinguish between (a), rules like UnionR*, and rules like transitivity? [1] Fabian Muehlboeck and Ross Tate. 2018. Empowering union and intersection types with integrated subtyping. Proc. ACM Program. Lang. 2, OOPSLA, Article 112 (October 2018), 29 pages. DOI: https://doi.org/10.1145/3276482 -- Kind regards, Julia From sergey.goncharov at fau.de Sat Jun 15 06:23:18 2019 From: sergey.goncharov at fau.de (Sergey Goncharov) Date: Sat, 15 Jun 2019 12:23:18 +0200 Subject: [TYPES] Variants and [Park or Scott] fixpoint Induction In-Reply-To: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> References: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> Message-ID: <10bef110-426b-efa0-a2da-c01efa71d81a@fau.de> Dear Peter, my understanding of the discrepancy you mention is like this. The expressive power of mu-calculus comes from Park induction *plus* negation. Since negation is in the language, we can dualize all the properties we can specify, in particular, define greatest fixpoints dually to the least ones, for which the dual Park rule gfp(F) => S iff Exists I. FI => I & I => S is derivable. This allows us to deal with safety since safety is dual to liveness (and liveness in mu-calculus is expressed via least fixpoints). Moreover, we can combine two kinds of fixpoints to express properties like fairness. Now, when it comes to the comparison between Hoare logic and mu-calculus, one has to be careful, because Hoare logic is exogenous (programs are explicit) and mu-calculus is endogenous (the underlying model represents the program of interest). Connecting Hoare logic with mu-calculus can be done via a calculus of weakest preconditions. This involves some curious dualization, because if we compute the semantics of a while-loop as a least fixpoint then the corresponding precondition involves a greatest fixpoint. So, connecting mu-calculus and Hoare logic introduces some confusion of fixpoints, but it does not change the fact that in mu-calculus we can freely use both kinds of fixpoints arbitrarily, while in Hoare logic we essentially have a global choice between weakest preconditions for total correctness and weakest (liberal) preconditions for partial correctness. Computationally the former are completely different kind of properties (co-r.e.) in comparison to the latter (r.e.) in the arithmetical hierarchy. I understand that the fact that mu-calculus is complete is essentially because it is complete wrt to a totality of over-abstracted programs, and so it is not affected by the above decidability issues. But in Hoare logic we deal with concrete programs for which the problem of checking partial correctness and the problem of checking total correctness are entirely asymmetric. In summary, I think your slogan "fixpoint induction is not good for termination? is correct because termination is not an inductive property and there is no conflict with the completeness of mu-calculus. Best, Sergey On 14/06/2019 16:23, O'Hearn, Peter wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > Two methods of reasoning about loops are provided by variants and by (Park or Scott) fixpoint induction. Is there a known relation or non-relation between them? My intuition is that fixpoint induction is not suitable for termination or liveness properties, but I am unsure whether this intuition is correct. > > The Hoare rule for total correctness of while loops using variants is well explained in the wikipedia article: > https://en.wikipedia.org/wiki/Hoare_logic#While_rule_for_total_correctness > There, you make sure a quantity in a well-founded set decreases on each loop iteration. > > Here is Park induction: > > lfp(F) <= S iff Exists I. FI <= I & I <= S > > If you think of S as ?spec? and I as ?invariant?, then this can form the basis for reasoning about safety properties (as explained by Cousot here) > http://web.mit.edu/16.399/www/lecture_11-b-fixpoints1/Cousot_MIT_2005_Course_11b_4-1.pdf > > I am a bit worried that my intuition "fixpoint induction is not good for termination? might have some holes in it. In particular, Park induction is used in a known complete proof theory for modal mu-calculus > https://eprints.illc.uva.nl/569/1/PP-2016-33.text.pdf > and that logic is notable for being able to express liveness properties. > > I asked a few experts who did not know a way to answer my question above, which is why I am posting it more widely here. In particular, if there is an explanation of how/why fixpoint induciton could be good for reasoning about (say) liveness or termination properties of while loops, I?f be glad to hear about it. > > Thanks! > > Peter O'Hearn > > > > > > > > > > > > -- PD Dr. Sergey Goncharov, Akademischer Oberrat FAU Erlangen-N?rnberg Phone: +49-91-3185-64031 Chair for TCS Fax: +49-91-3185-64055 Martenstra?e 3 Email: Sergey.Goncharov at cs.fau.de D-91058 Erlangen Web: http://www8.cs.fau.de/~sergey From streicher at mathematik.tu-darmstadt.de Sat Jun 15 06:46:54 2019 From: streicher at mathematik.tu-darmstadt.de (Thomas Streicher) Date: Sat, 15 Jun 2019 12:46:54 +0200 Subject: [TYPES] Variants and [Park or Scott] fixpoint Induction In-Reply-To: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> References: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> Message-ID: <20190615104654.GB7888@mathematik.tu-darmstadt.de> > Two methods of reasoning about loops are provided by variants and by (Park or Scott) fixpoint induction. Is there a known relation or non-relation between them? My intuition is that fixpoint induction is not suitable for termination or liveness properties, but I am unsure whether this intuition is correct. For fixpoint (i.e. Scott) induction you have to assume that the predicate under consideration contains bottom and is closed under sups of increasing chains. Since it contains bottom such a predicate can't express any kind of termination property. Termination properties of programs P = fix(Phi) are usually proved by induction over data types from P = Phi(P), i.e. the defining equation for P. Park induction also only allows one to prove that mu(Phi) is contained in some f \geq Phi(f), i.e. also can prove only downward closed properties. That explains why logic of programs which are not of domain-theoretic kind, i.e. don't allow one to express infomation ordering, are successful for proving total correctness properties. But I think that proving negative properties like nontermination is also very important and for this reason domain theory is very important. Thomas From p.b.levy at cs.bham.ac.uk Sat Jun 15 07:45:40 2019 From: p.b.levy at cs.bham.ac.uk (Paul Blain Levy) Date: Sat, 15 Jun 2019 12:45:40 +0100 Subject: [TYPES] Theory of Particular Sets In-Reply-To: References: Message-ID: Hi, I've posted a paper about a new set theory that is somewhat influenced by dependent type theory. Any comments are welcome. Paul -- https://arxiv.org/abs/1905.02718 A Theory of Particular Sets ZFC has sentences that quantify over all sets or all ordinals, without restriction. Some have argued that sentences of this kind lack a determinate meaning. We propose a set theory called TOPS, using Natural Deduction, that avoids this problem by speaking only about particular sets. From p.ohearn at ucl.ac.uk Sat Jun 15 08:12:19 2019 From: p.ohearn at ucl.ac.uk (O'Hearn, Peter) Date: Sat, 15 Jun 2019 12:12:19 +0000 Subject: [TYPES] Variants and [Park or Scott] fixpoint Induction In-Reply-To: <20190615104654.GB7888@mathematik.tu-darmstadt.de> References: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> <20190615104654.GB7888@mathematik.tu-darmstadt.de> Message-ID: <7B35AB9B-2479-4483-A117-DF4F9849D582@ucl.ac.uk> On 15 Jun 2019, at 11:46, Thomas Streicher > wrote: Two methods of reasoning about loops are provided by variants and by (Park or Scott) fixpoint induction. Is there a known relation or non-relation between them? My intuition is that fixpoint induction is not suitable for termination or liveness properties, but I am unsure whether this intuition is correct. For fixpoint (i.e. Scott) induction you have to assume that the predicate under consideration contains bottom and is closed under sups of increasing chains. Since it contains bottom such a predicate can't express any kind of termination property. Thomas, this was precisely what I meant. Termination is not ?admissible for fixed-point induction?. And other techniques (e.g., based on ranking functions) are commonly used to prove termination. The only thing that gives me pause is the existence of proof theories for temporal/modal formalisms (e.g., mu-calculus) that are based on Park induciton and which are complete, and yet the formalisms can express liveness properties too. Possibly games are being played with greatest FP?s, or perhaps they cannot actually prove the kinds of properties one would want in order to show the termination argument in Hoare Logic for total correctness. I am not sure. But thanks for your answer. Peter From fp at cs.cmu.edu Sat Jun 15 09:31:50 2019 From: fp at cs.cmu.edu (Frank Pfenning) Date: Sat, 15 Jun 2019 09:31:50 -0400 Subject: [TYPES] What exactly should we call syntax-directed inference rules? In-Reply-To: References: Message-ID: In the proof theory tradition, rules like those you quote for unions are called *analytic*. It means not only that there are a finite number of possibilities, but they are composed of subformulas of the conclusion. So, like Pierce, I would tend to reserve *syntax-directed* for the stronger property that the rule selection is determined unambiguously by the syntax, since we already have a perfectly good term for the weaker property. - Frank On Fri, Jun 14, 2019 at 7:16 PM Julia Belyakova wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear All, > > It seems that there is a slight inconsistency in what kind of inference > rules are called "syntax-directed". Let's take subtyping rules as an > example. > > (a) In TAPL, syntax-directed means that for each pair of types T and S, > there is exactly one rule with the conclusion that matches T <: S. > (b) In some papers, e.g. [1], rules are considered syntax-directed if for > each pair T and S there is a finite number of rules matching T <: S and > each of them has a finite number of (computable) premises. > > The transitivity rule > > T1 <: T2 T2 <: T3 > ---------------------- > T1 <: T3 > > is clearly bad, and it is rejected by both definitions: > (a) The conclusion T1 <: T3 overlaps with any other subtyping rule. > (b) It is not clear how to compute T2 and there can be infinitely many of > them. > > But what about subtyping of union types? > > T <: S1 T <: S2 > ------------- UnionR1 ------------ UnionR2 > T <: S1?S2 T <: S1?S2 > > According to (a), such rules are not syntax-directed. According to (b), > they are. > > The UnionR* rules are definitely worse than syntax-directed rules in the > sense of (a) because for a triple of types T, S1, S2 we have two options, > and one of them might work while the other does not. Algorithmically, this > is the source of backtracking. > However, these rules are better than the transitivity rule because their > premises are uniquely defined by syntax. > > Which terminology is more generally accepted, (a) or (b)? And is there a > suitable terminology to distinguish between (a), rules like UnionR*, and > rules like transitivity? > > [1] Fabian Muehlboeck and Ross Tate. 2018. Empowering union and > intersection types with integrated subtyping. Proc. ACM Program. Lang. 2, > OOPSLA, Article 112 (October 2018), 29 pages. DOI: > https://doi.org/10.1145/3276482 > > -- > Kind regards, Julia > -- Frank Pfenning, Professor Computer Science Department Carnegie Mellon University Pittsburgh, PA 15213-3891 http://www.cs.cmu.edu/~fp +1 412 268-6343 GHC 6017 From sherman at csail.mit.edu Sat Jun 15 11:11:41 2019 From: sherman at csail.mit.edu (Ben Sherman) Date: Sat, 15 Jun 2019 11:11:41 -0400 Subject: [TYPES] Variants and [Park or Scott] fixpoint Induction In-Reply-To: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> References: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> Message-ID: <05CFCE8E-719D-4DF6-B91F-216505A7C8B5@csail.mit.edu> I think, if I?m interpreting your question correctly, that your intuition is correct. Park?s upper fixpoint induction principle that you state allows you to prove upper bounds on lfp(F), which I interpret as safety properties, whereas I interpret liveness properties as lower bounds on lfp(F). Looking at the paper on the modal mu-calculus, I would think that the liveness properties arise not from the Park induction rule schema, but rather the pre-fixpoint axiom schema, which says (reinterpreting in your syntax the first equation on page 4): F(lfp(F)) <= lfp(F) It is this axiom schema that allows one to prove lower bounds on lfp(F), which I interpret as liveness properties. Very informally, I believe that a loop variant ought to correspond to the number of times the above axiom must be applied. (But I?m not entirely confident about this!) Ben > On Jun 14, 2019, at 11:23 AM, O'Hearn, Peter wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > Two methods of reasoning about loops are provided by variants and by (Park or Scott) fixpoint induction. Is there a known relation or non-relation between them? My intuition is that fixpoint induction is not suitable for termination or liveness properties, but I am unsure whether this intuition is correct. > > The Hoare rule for total correctness of while loops using variants is well explained in the wikipedia article: > https://en.wikipedia.org/wiki/Hoare_logic#While_rule_for_total_correctness > There, you make sure a quantity in a well-founded set decreases on each loop iteration. > > Here is Park induction: > > lfp(F) <= S iff Exists I. FI <= I & I <= S > > If you think of S as ?spec? and I as ?invariant?, then this can form the basis for reasoning about safety properties (as explained by Cousot here) > http://web.mit.edu/16.399/www/lecture_11-b-fixpoints1/Cousot_MIT_2005_Course_11b_4-1.pdf > > I am a bit worried that my intuition "fixpoint induction is not good for termination? might have some holes in it. In particular, Park induction is used in a known complete proof theory for modal mu-calculus > https://eprints.illc.uva.nl/569/1/PP-2016-33.text.pdf > and that logic is notable for being able to express liveness properties. > > I asked a few experts who did not know a way to answer my question above, which is why I am posting it more widely here. In particular, if there is an explanation of how/why fixpoint induciton could be good for reasoning about (say) liveness or termination properties of while loops, I?f be glad to hear about it. > > Thanks! > > Peter O'Hearn > > > > > > > > > > > > From tadeusz.litak at gmail.com Sat Jun 15 13:05:15 2019 From: tadeusz.litak at gmail.com (Tadeusz Litak) Date: Sat, 15 Jun 2019 19:05:15 +0200 Subject: [TYPES] Variants and [Park or Scott] fixpoint Induction In-Reply-To: <10bef110-426b-efa0-a2da-c01efa71d81a@fau.de> References: <83B29CB1-1BE9-4034-AFDC-465BA8424607@ucl.ac.uk> <10bef110-426b-efa0-a2da-c01efa71d81a@fau.de> Message-ID: <8bd64435-8196-347f-ea5e-6e1f7cda7836@gmail.com> On 15.06.19 12:23, Sergey Goncharov wrote: > > in Hoare logic we essentially have a global choice between weakest > preconditions for total correctness and weakest (liberal) > preconditions for partial correctness. Computationally the former are > completely different kind of properties (co-r.e.) in comparison to the > latter (r.e.) in the arithmetical hierarchy. This seems to depend on the programming language you are talking about. For example, this is what Clarke, German and Halpern write in the conclusion of their POPL 1982 paper "On effective axiomatizations of Hoare logics": > This leads us to our last point: the relationship between partial > correctness and termination, > and our ability to find good axiom systems for complicated programming > languages. One conclusion we can draw is that under the assumption > that the halting problem is decidable for finite interpretations, > partial correctness and termination seem to have essentially the same > complexity. However, for more complicated deterministic programming > languages such as those discussed in [Cl76/79] which do not have a > decidable halting problem for finite interpretations,? termination > assertions, and hence total correctness assertions, are effectively > axiomatizable, while partial correctness assertions are not. This > suggests the use of a total correctness proof system which, unlike > most currently available, does not require the establishment of > partial correctness as an essential first step. From julbinb at gmail.com Mon Jun 17 08:56:30 2019 From: julbinb at gmail.com (Julia Belyakova) Date: Mon, 17 Jun 2019 08:56:30 -0400 Subject: [TYPES] What exactly should we call syntax-directed inference rules? In-Reply-To: References: Message-ID: Thanks, Frank! That's exactly it. I see there are various papers on logic, proof search, analytic vs non-analytic proof systems, which use the term but do not define it. Your 1984 paper calls analytic "resolution or mating proof systems". I wonder if there is there a classic paper/book I can refer to for the definition? -- Kind regards, Julia ??, 15 ???. 2019 ?. ? 09:32, Frank Pfenning : > In the proof theory tradition, rules like those you quote for unions are > called *analytic*. It means not only that there are a finite number of > possibilities, > but they are composed of subformulas of the conclusion. So, like Pierce, > I would tend to reserve *syntax-directed* for the stronger property that > the > rule selection is determined unambiguously by the syntax, since we already > have a > perfectly good term for the weaker property. > > - Frank > > On Fri, Jun 14, 2019 at 7:16 PM Julia Belyakova wrote: > >> [ The Types Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> >> Dear All, >> >> It seems that there is a slight inconsistency in what kind of inference >> rules are called "syntax-directed". Let's take subtyping rules as an >> example. >> >> (a) In TAPL, syntax-directed means that for each pair of types T and S, >> there is exactly one rule with the conclusion that matches T <: S. >> (b) In some papers, e.g. [1], rules are considered syntax-directed if for >> each pair T and S there is a finite number of rules matching T <: S and >> each of them has a finite number of (computable) premises. >> >> The transitivity rule >> >> T1 <: T2 T2 <: T3 >> ---------------------- >> T1 <: T3 >> >> is clearly bad, and it is rejected by both definitions: >> (a) The conclusion T1 <: T3 overlaps with any other subtyping rule. >> (b) It is not clear how to compute T2 and there can be infinitely many of >> them. >> >> But what about subtyping of union types? >> >> T <: S1 T <: S2 >> ------------- UnionR1 ------------ UnionR2 >> T <: S1?S2 T <: S1?S2 >> >> According to (a), such rules are not syntax-directed. According to (b), >> they are. >> >> The UnionR* rules are definitely worse than syntax-directed rules in the >> sense of (a) because for a triple of types T, S1, S2 we have two options, >> and one of them might work while the other does not. Algorithmically, this >> is the source of backtracking. >> However, these rules are better than the transitivity rule because their >> premises are uniquely defined by syntax. >> >> Which terminology is more generally accepted, (a) or (b)? And is there a >> suitable terminology to distinguish between (a), rules like UnionR*, and >> rules like transitivity? >> >> [1] Fabian Muehlboeck and Ross Tate. 2018. Empowering union and >> intersection types with integrated subtyping. Proc. ACM Program. Lang. 2, >> OOPSLA, Article 112 (October 2018), 29 pages. DOI: >> https://doi.org/10.1145/3276482 >> >> -- >> Kind regards, Julia >> > > > -- > Frank Pfenning, Professor > Computer Science Department > Carnegie Mellon University > Pittsburgh, PA 15213-3891 > > http://www.cs.cmu.edu/~fp > +1 412 268-6343 > GHC 6017 > > From fp at cs.cmu.edu Mon Jun 17 09:37:20 2019 From: fp at cs.cmu.edu (Frank Pfenning) Date: Mon, 17 Jun 2019 06:37:20 -0700 Subject: [TYPES] What exactly should we call syntax-directed inference rules? In-Reply-To: References: Message-ID: I am not a historian, but the classic paper I would cite is Gentzen's 1935 paper that introduced both natural deduction and the sequent calculus. Cut-free sequent proofs have the subformula property and are therefore analytic, although I don't think that Gentzen actually used the term. In natural deduction, I think Prawitz 1965 was the first one to explicitly characterizes the normal forms (which are the analytic natural deductions). The Stanford Encyclopedia of Philosophy (*The Analytic/Synthetic Distinction* ) credits Kant 1781 who defines the term and elaborates. Martin-L?f 1994 elaborates on what this should mean in type theory in his paper *Analytic and Synthetic Judgements in Type Theory .* - Frank On Mon, Jun 17, 2019 at 6:03 AM Julia Belyakova wrote: > Thanks, Frank! That's exactly it. > > I see there are various papers on logic, proof search, analytic vs > non-analytic proof systems, which use the term but do not define it. Your > 1984 paper calls analytic "resolution or mating proof systems". > I wonder if there is there a classic paper/book I can refer to for the > definition? > > -- > Kind regards, Julia > > > ??, 15 ???. 2019 ?. ? 09:32, Frank Pfenning : > >> In the proof theory tradition, rules like those you quote for unions are >> called *analytic*. It means not only that there are a finite number of >> possibilities, >> but they are composed of subformulas of the conclusion. So, like Pierce, >> I would tend to reserve *syntax-directed* for the stronger property that >> the >> rule selection is determined unambiguously by the syntax, since we >> already have a >> perfectly good term for the weaker property. >> >> - Frank >> >> On Fri, Jun 14, 2019 at 7:16 PM Julia Belyakova >> wrote: >> >>> [ The Types Forum, >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >>> >>> Dear All, >>> >>> It seems that there is a slight inconsistency in what kind of inference >>> rules are called "syntax-directed". Let's take subtyping rules as an >>> example. >>> >>> (a) In TAPL, syntax-directed means that for each pair of types T and S, >>> there is exactly one rule with the conclusion that matches T <: S. >>> (b) In some papers, e.g. [1], rules are considered syntax-directed if for >>> each pair T and S there is a finite number of rules matching T <: S and >>> each of them has a finite number of (computable) premises. >>> >>> The transitivity rule >>> >>> T1 <: T2 T2 <: T3 >>> ---------------------- >>> T1 <: T3 >>> >>> is clearly bad, and it is rejected by both definitions: >>> (a) The conclusion T1 <: T3 overlaps with any other subtyping rule. >>> (b) It is not clear how to compute T2 and there can be infinitely many of >>> them. >>> >>> But what about subtyping of union types? >>> >>> T <: S1 T <: S2 >>> ------------- UnionR1 ------------ UnionR2 >>> T <: S1?S2 T <: S1?S2 >>> >>> According to (a), such rules are not syntax-directed. According to (b), >>> they are. >>> >>> The UnionR* rules are definitely worse than syntax-directed rules in the >>> sense of (a) because for a triple of types T, S1, S2 we have two options, >>> and one of them might work while the other does not. Algorithmically, >>> this >>> is the source of backtracking. >>> However, these rules are better than the transitivity rule because their >>> premises are uniquely defined by syntax. >>> >>> Which terminology is more generally accepted, (a) or (b)? And is there a >>> suitable terminology to distinguish between (a), rules like UnionR*, and >>> rules like transitivity? >>> >>> [1] Fabian Muehlboeck and Ross Tate. 2018. Empowering union and >>> intersection types with integrated subtyping. Proc. ACM Program. Lang. 2, >>> OOPSLA, Article 112 (October 2018), 29 pages. DOI: >>> https://doi.org/10.1145/3276482 >>> >>> -- >>> Kind regards, Julia >>> >> >> >> -- >> Frank Pfenning, Professor >> Computer Science Department >> Carnegie Mellon University >> Pittsburgh, PA 15213-3891 >> >> http://www.cs.cmu.edu/~fp >> +1 412 268-6343 >> GHC 6017 >> >> -- Frank Pfenning, Professor Computer Science Department Carnegie Mellon University Pittsburgh, PA 15213-3891 http://www.cs.cmu.edu/~fp +1 412 268-6343 GHC 6017 From julbinb at gmail.com Mon Jun 17 18:38:01 2019 From: julbinb at gmail.com (Julia Belyakova) Date: Mon, 17 Jun 2019 18:38:01 -0400 Subject: [TYPES] What exactly should we call syntax-directed inference rules? In-Reply-To: <001601d52552$981f6d00$c85e4700$@cs.cornell.edu> References: <001601d52552$981f6d00$c85e4700$@cs.cornell.edu> Message-ID: Dear Fabian and Ross, Thanks for the clarification! Also, I wonder if the rule NF(t) <: t' ------------- t <: t' can be considered analytic. Strictly speaking, a normal form is not a subformula, though the premise of the rule is completely determined by the conclusion, which seems to be in the analytic spirit. -- Kind regards, Julia ??, 17 ???. 2019 ?. ? 17:21, Fabian Muehlboeck : > Hi all, > A complication is that our system neither requires nor produces > syntax-directed or analytic systems (according to the definitions you have > provided). Julia, from what we know of what you are working on, it is > likely your system also satisfies neither definition. The problem is that > the types in the premises are not necessarily subterms of the types (or > their instances) in the conclusion, a point of flexibility that seems very > important for achieving the kinds of expressiveness we (both) want. All we > assume/provide is that any conclusion has a (computable) finite number of > applicable rules (with metavariables instantiated) and that those rules > have a (computable) finite number of premises. Importantly, not having the > subformula property means that in general proof search is not necessarily > terminating (even with cycle detection), so we need another property of the > system to ensure decidability (the same goes for non-analytic > syntax-directed systems, as TAPL notes), which we called well-foundedness. > It would be great to have standard terms for these notions. Unfortunately, > we could not find them ourselves, and so we generalized existing terms. > Frank (or anyone else on the list), do you happen to know the precise terms > that describe these two concepts that make proof search a decision > procedure? > > Thanks > > Fabian and Ross > > > -----Original Message----- > From: Types-list On Behalf Of > Frank Pfenning > Sent: Monday, June 17, 2019 09:37 > To: Julia Belyakova > Cc: Types list > Subject: Re: [TYPES] What exactly should we call syntax-directed inference > rules? > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > I am not a historian, but the classic paper I would cite is Gentzen's 1935 > paper that introduced both natural deduction and the sequent calculus. > Cut-free sequent proofs have the subformula property and are therefore > analytic, although I don't think that Gentzen actually used the term. > In natural deduction, I think Prawitz 1965 was the first one to explicitly > characterizes the normal forms (which are the analytic natural deductions). > > The Stanford Encyclopedia of Philosophy (*The Analytic/Synthetic > Distinction* ) > credits Kant 1781 who defines the term and elaborates. Martin-L?f 1994 > elaborates on what this should mean in type theory in his paper *Analytic > and Synthetic Judgements in Type Theory > .* > > - Frank > > On Mon, Jun 17, 2019 at 6:03 AM Julia Belyakova wrote: > > > Thanks, Frank! That's exactly it. > > > > I see there are various papers on logic, proof search, analytic vs > > non-analytic proof systems, which use the term but do not define it. > > Your > > 1984 paper calls analytic "resolution or mating proof systems". > > I wonder if there is there a classic paper/book I can refer to for the > > definition? > > > > -- > > Kind regards, Julia > > > > > > ??, 15 ???. 2019 ?. ? 09:32, Frank Pfenning : > > > >> In the proof theory tradition, rules like those you quote for unions > >> are called *analytic*. It means not only that there are a finite > >> number of possibilities, but they are composed of subformulas of the > >> conclusion. So, like Pierce, I would tend to reserve > >> *syntax-directed* for the stronger property that the rule selection > >> is determined unambiguously by the syntax, since we already have a > >> perfectly good term for the weaker property. > >> > >> - Frank > >> > >> On Fri, Jun 14, 2019 at 7:16 PM Julia Belyakova > >> wrote: > >> > >>> [ The Types Forum, > >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > >>> > >>> Dear All, > >>> > >>> It seems that there is a slight inconsistency in what kind of > >>> inference rules are called "syntax-directed". Let's take subtyping > >>> rules as an example. > >>> > >>> (a) In TAPL, syntax-directed means that for each pair of types T and > >>> S, there is exactly one rule with the conclusion that matches T <: S. > >>> (b) In some papers, e.g. [1], rules are considered syntax-directed > >>> if for each pair T and S there is a finite number of rules matching > >>> T <: S and each of them has a finite number of (computable) premises. > >>> > >>> The transitivity rule > >>> > >>> T1 <: T2 T2 <: T3 > >>> ---------------------- > >>> T1 <: T3 > >>> > >>> is clearly bad, and it is rejected by both definitions: > >>> (a) The conclusion T1 <: T3 overlaps with any other subtyping rule. > >>> (b) It is not clear how to compute T2 and there can be infinitely > >>> many of them. > >>> > >>> But what about subtyping of union types? > >>> > >>> T <: S1 T <: S2 > >>> ------------- UnionR1 ------------ UnionR2 > >>> T <: S1?S2 T <: S1?S2 > >>> > >>> According to (a), such rules are not syntax-directed. According to > >>> (b), they are. > >>> > >>> The UnionR* rules are definitely worse than syntax-directed rules in > >>> the sense of (a) because for a triple of types T, S1, S2 we have two > >>> options, and one of them might work while the other does not. > >>> Algorithmically, this is the source of backtracking. > >>> However, these rules are better than the transitivity rule because > >>> their premises are uniquely defined by syntax. > >>> > >>> Which terminology is more generally accepted, (a) or (b)? And is > >>> there a suitable terminology to distinguish between (a), rules like > >>> UnionR*, and rules like transitivity? > >>> > >>> [1] Fabian Muehlboeck and Ross Tate. 2018. Empowering union and > >>> intersection types with integrated subtyping. Proc. ACM Program. > >>> Lang. 2, OOPSLA, Article 112 (October 2018), 29 pages. DOI: > >>> https://doi.org/10.1145/3276482 > >>> > >>> -- > >>> Kind regards, Julia > >>> > >> > >> > >> -- > >> Frank Pfenning, Professor > >> Computer Science Department > >> Carnegie Mellon University > >> Pittsburgh, PA 15213-3891 > >> > >> http://www.cs.cmu.edu/~fp > >> +1 412 268-6343 > >> GHC 6017 > >> > >> > > -- > Frank Pfenning, Professor > Computer Science Department > Carnegie Mellon University > Pittsburgh, PA 15213-3891 > > http://www.cs.cmu.edu/~fp > +1 412 268-6343 > GHC 6017 > > From moez at cs.rice.edu Thu Jun 27 05:28:01 2019 From: moez at cs.rice.edu (moez at cs.rice.edu) Date: Thu, 27 Jun 2019 09:28:01 +0000 (UTC) Subject: [TYPES] An order-theoretic approach to OO generics Message-ID: Dear TYPEers, The progress of OOP languages such as Java, C#, Scala, and C++ has been hindered by the complexity of their type systems. Features of generics in OOP languages such as: (1) variance annotations (of which Java wildcards are an instance), (2) (Java) erasure, where type arguments of a parameterized type get dropped/ignored, and (3) F-bounded polymorphism/generics, where a type variable occurs in its own bound, such as in the Java class declaration class Enum>{}, are particularly notable for their contribution to the complexity of OO type systems (as a check of the latest versions of language specifications - or of the Java generics FAQ - reveals, these three particular features make the specification of type systems of these languages overly unwieldy and complex). This complexity makes the full understanding of the type systems of these mainstream OO languages amenable only to the most advanced of OO software developers. Building on the simplicity and intuitiveness of graph theory, order theory, and *basic* category theory, in recent work I've tried to build a simple model of generics in nominally-typed OOP languages (such as the ones mentioned above) using an approach that is mainly and mostly order-theoretic in its essence and nature. I presented the details of this approach in multiple research articles that I posted to arXiv. To summarize the approach though,? recently I presented an abridged outline of the whole approach (as developed so far) in an article available at: http://arxiv.org/abs/1906.11197 I am posting about this outline article to the TYPES mailing list because, first, I'd like to thank all those (mathematicians, PL researchers, language designers, and OO software developers) who gave me their precious feedback on the approach while its development was in its infancy.? That includes thanking in particular David Spivak, Robert Cartwright, John Greiner, Daniel Smith, Larry Paulson, Nathan Carter, and Kariem ElKoush, as well as plenty of the anonymous reviewers of conferences and workshops I submitted some of my work to.? Any shortcomings or errors found in the approach as developed so far or in its presentation should be blamed on me and no one else. And, since the approach has matured a bit, I am now seeking further feedback on the approach from the wider community. Next, given the limited financial and human resources available to me, I am also seeking collaborations with interested and qualified members of the community, which can help in speeding up the development of the approach and in finalizing it sooner. Finally, if you happen to read the outline paper then please make sure that (even if you can provide me with no material support) you do not keep your opinions about the approach to yourself.? All kinds of feedback on the approach --- including constructive, well-justified and respectful criticism --- are more than welcome. Cheers, -Moez PS: Pending finding funding for my trip, I plan to be in Oxford in mid July to attend ACT 2019 and to present a poster there on this work, under the title 'Modeling OO Generics: An Order- and Category-Theoretic Approach'.? Assuming I make it to Oxford, those attending ACT 2019 and interested in a chat should feel free to drop by. From monnier at iro.umontreal.ca Fri Jul 5 18:08:16 2019 From: monnier at iro.umontreal.ca (Stefan Monnier) Date: Fri, 05 Jul 2019 18:08:16 -0400 Subject: [TYPES] Example uses of impredicativity Message-ID: Hi, I was recently reminded that not only I only understand impredicativity in a very superficial way, but on top of that, I don't know of many uses of it. Basically, I know of its use in Church encodings, in Shao et. al.'s ?? (from Type-safe Certified Binaries, which I also used in my "swiss coercion"), as well as in type-preserving closure conversion, but that's pretty much it. When I started to look for examples in Coq, I soon realized that many Coq libraries have definitions whose universe level is affected by impredicativity but it's a lot more work to discover if the same development would work as well without impredicativity. Is there some place where I could find a kind of "collection" of developments/techniques which rely on impredicativity? Short of that, I'd be happy to hear from examples which I could assemble to make such a collection. Stefan From wjb at williamjbowman.com Sat Jul 6 02:16:30 2019 From: wjb at williamjbowman.com (William J. Bowman) Date: Fri, 5 Jul 2019 23:16:30 -0700 Subject: [TYPES] Example uses of impredicativity In-Reply-To: References: Message-ID: Stefan, The type preserving CPS translation with answer type polymorphism requires impredicativity. C?dille uses impredicativity heavily, so that might be a better place to look than Coq. Although I guess that's more examples of (things like) Church encodings. (For that matter, the CPS translation is too.) -- Sent from my phoneamajig > On Jul 5, 2019, at 22:33, William J. Bowman wrote: > > Stefan, > > The type preserving CPS translation with answer type polymorphism requires impredicativity. From m.escardo at cs.bham.ac.uk Sat Jul 6 03:44:43 2019 From: m.escardo at cs.bham.ac.uk (Martin Escardo) Date: Sat, 6 Jul 2019 08:44:43 +0100 Subject: [TYPES] Example uses of impredicativity In-Reply-To: References: Message-ID: <31108453-c214-06ff-b507-ec5fe58a660c@cs.bham.ac.uk> One example is the powerset P X of a type X, defined as the type X -> Prop of functions into Prop. Without the impredicativity of Prop, you are not able to construct a function P(P X) -> P X that gives the union of a set of sets. Also Prop itself wouldn't be a complete lattice with a join function P Prop -> Prop in the absence of impredictivity. Martin On 05/07/2019 23:08, monnier at iro.umontreal.ca wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Hi, > > I was recently reminded that not only I only understand impredicativity > in a very superficial way, but on top of that, I don't know of many > uses of it. > > Basically, I know of its use in Church encodings, in Shao et. al.'s ?? > (from Type-safe Certified Binaries, which I also used in my "swiss > coercion"), as well as in type-preserving closure conversion, but that's > pretty much it. > > When I started to look for examples in Coq, I soon realized that many > Coq libraries have definitions whose universe level is affected by > impredicativity but it's a lot more work to discover if the same > development would work as well without impredicativity. > > Is there some place where I could find a kind of "collection" of > developments/techniques which rely on impredicativity? > > Short of that, I'd be happy to hear from examples which > I could assemble to make such a collection. > > > Stefan > -- Martin Escardo http://www.cs.bham.ac.uk/~mhe From bcpierce at cis.upenn.edu Mon Jul 8 12:57:42 2019 From: bcpierce at cis.upenn.edu (Benjamin C. Pierce) Date: Mon, 8 Jul 2019 12:57:42 -0400 Subject: [TYPES] Example uses of impredicativity In-Reply-To: References: Message-ID: <46B41A05-3EC9-461D-A131-73D5C2B6E7C9@cis.upenn.edu> Impredicativity is also useful in type-theoretic encodings of objects. Objects with method interface M are represented as elements of Obj(M) = Exists X, X * (X -> M(X)) i.e., as pairs of a ?hidden state? X and a transformation on that state yielding a result of shape M. For example, if M(X) = { push: Nat -> X, pop: Nat * X } then Obj(M) can be viewed as a type of stack objects. And since one natural way to represent a non-empty stack is as a pair of a number and another stack, we need the existential quantifier to range over a collection of types that includes Nat * Obj(M). (This is basically a generalization of the typed closure encoding, though they were actually discovered in the other order!) Best, - Benjamin > On Jul 5, 2019, at 6:08 PM, Stefan Monnier wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Hi, > > I was recently reminded that not only I only understand impredicativity > in a very superficial way, but on top of that, I don't know of many > uses of it. > > Basically, I know of its use in Church encodings, in Shao et. al.'s ?? > (from Type-safe Certified Binaries, which I also used in my "swiss > coercion"), as well as in type-preserving closure conversion, but that's > pretty much it. > > When I started to look for examples in Coq, I soon realized that many > Coq libraries have definitions whose universe level is affected by > impredicativity but it's a lot more work to discover if the same > development would work as well without impredicativity. > > Is there some place where I could find a kind of "collection" of > developments/techniques which rely on impredicativity? > > Short of that, I'd be happy to hear from examples which > I could assemble to make such a collection. > > > Stefan > > From beta.ziliani at gmail.com Mon Jul 8 18:07:12 2019 From: beta.ziliani at gmail.com (Beta Ziliani) Date: Mon, 8 Jul 2019 19:07:12 -0300 Subject: [TYPES] Example uses of impredicativity In-Reply-To: References: Message-ID: Hi Stefan, In Mtac we use impredicativity to code a fixpoint for Coq?s meta-programs. The details can be found in [1]. Roughly, the idea behind Mtac is to have an inductive datatype M : Type -> Prop with primitives for meta-programming in Coq. For instance, you have monadic operations (ret : forall A:Type, A -> M A, bind : forall A B : Type, M A -> (A -> M B) -> M B) together with meta-operations, like meta-variable creation (evar : forall A: Type, M A), etc. It also has an unbounded fixpoint. However, the typical type for fixpoints doesn?t work with Coq, as it requires a negative occurrence of M: mfix : forall (A: Type) (P: A -> Type), ((forall x: A, *M (P x)*) -> (forall x: A, M (P x))) -> (forall x: A, M (P x)) (note the first M). For this reason, we encode this using the ?Mendler style?, adding a parameter S: mfix : forall (A: Type) (P: A -> Type) *(S: Type -> ????)*, *(forall x: A, S (P x) -> M (P x)) ->* ((forall x: A, S (P x)) -> (forall x: A, S (P x))) -> (forall x: A, M (P x)) (The ???? will become clear in a sec.) Now we avoid the negative occurrence on M, but we have a new condition relating S and M: forall x: A, S (P x) -> M (P x). This condition will be instantiated with the identity. But if S and M where Types instead of Props, that would lead to a circularity in the universes. Therefore, we need ???? to be a Prop, the same as M?s co-domain. Best, Beta [1] Mtac: A Monad for Typed Tactic Programming in Coq. B. Ziliani, D. Dreyer, N. R. Krishnaswami, A. Nanevski, V. Vafeiadis. In Journal of Functional Programming (JFP), Volume 25, August 2015. On Sat, Jul 6, 2019 at 2:08 AM Stefan Monnier wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Hi, > > I was recently reminded that not only I only understand impredicativity > in a very superficial way, but on top of that, I don't know of many > uses of it. > > Basically, I know of its use in Church encodings, in Shao et. al.'s ?? > (from Type-safe Certified Binaries, which I also used in my "swiss > coercion"), as well as in type-preserving closure conversion, but that's > pretty much it. > > When I started to look for examples in Coq, I soon realized that many > Coq libraries have definitions whose universe level is affected by > impredicativity but it's a lot more work to discover if the same > development would work as well without impredicativity. > > Is there some place where I could find a kind of "collection" of > developments/techniques which rely on impredicativity? > > Short of that, I'd be happy to hear from examples which > I could assemble to make such a collection. > > > Stefan > > From matthieu.sozeau at inria.fr Wed Jul 24 15:12:45 2019 From: matthieu.sozeau at inria.fr (Matthieu Sozeau) Date: Wed, 24 Jul 2019 15:12:45 -0400 Subject: [TYPES] Example uses of impredicativity In-Reply-To: <31108453-c214-06ff-b507-ec5fe58a660c@cs.bham.ac.uk> References: <31108453-c214-06ff-b507-ec5fe58a660c@cs.bham.ac.uk> Message-ID: <1970C64A-C2BA-4CF4-AEEC-CC11BA92D68C@inria.fr> > Le 6 juil. 2019 ? 03:44, Martin Escardo a ?crit : > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > One example is the powerset P X of a type X, defined as the type X -> Prop of functions into Prop. Without the impredicativity of Prop, you are not able to construct a function P(P X) -> P X that gives the union of a set of sets. Also Prop itself wouldn't be a complete lattice with a join function P Prop -> Prop in the absence of impredictivity. ? and this complete lattice gives you a Tarski fixed point theorem, on which the Paco library for coinductive reasoning is built, making use of Mendler-style recursion (also requiring impredicativity). ? Matthieu > On 05/07/2019 23:08, monnier at iro.umontreal.ca wrote: >> [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> Hi, >> I was recently reminded that not only I only understand impredicativity >> in a very superficial way, but on top of that, I don't know of many >> uses of it. >> Basically, I know of its use in Church encodings, in Shao et. al.'s ?? >> (from Type-safe Certified Binaries, which I also used in my "swiss >> coercion"), as well as in type-preserving closure conversion, but that's >> pretty much it. >> When I started to look for examples in Coq, I soon realized that many >> Coq libraries have definitions whose universe level is affected by >> impredicativity but it's a lot more work to discover if the same >> development would work as well without impredicativity. >> Is there some place where I could find a kind of "collection" of >> developments/techniques which rely on impredicativity? >> Short of that, I'd be happy to hear from examples which >> I could assemble to make such a collection. >> Stefan > > -- > Martin Escardo > http://www.cs.bham.ac.uk/~mhe From moez at cs.rice.edu Tue Jul 30 02:26:50 2019 From: moez at cs.rice.edu (moez at cs.rice.edu) Date: Tue, 30 Jul 2019 06:26:50 +0000 (UTC) Subject: [TYPES] An order-theoretic approach to OO generics In-Reply-To: References: Message-ID: Dear all, Here's my ACT 2019 poster.? It summarizes the order-theoretic (and category-theoretic) approach to modeling generics. https://app.box.com/s/o3xv02sbic4uswh2700wlt46bbjjgnke Regards, -Moez From: moez at cs.rice.edu Sent: Thursday, June 27, 11:28 Subject: An order-theoretic approach to OO generics To: Types List Dear TYPEers, The progress of OOP languages such as Java, C#, Scala, and C++ has been hindered by the complexity of their type systems. Features of generics in OOP languages such as: (1) variance annotations (of which Java wildcards are an instance), (2) (Java) erasure, where type arguments of a parameterized type get dropped/ignored, and (3) F-bounded polymorphism/generics, where a type variable occurs in its own bound, such as in the Java class declaration class Enum>{}, are particularly notable for their contribution to the complexity of OO type systems (as a check of the latest versions of language specifications - or of the Java generics FAQ - reveals, these three particular features make the specification of type systems of these languages overly unwieldy and complex). This complexity makes the full understanding of the type systems of these mainstream OO languages amenable only to the most advanced of OO software developers. Building on the simplicity and intuitiveness of graph theory, order theory, and *basic* category theory, in recent work I've tried to build a simple model of generics in nominally-typed OOP languages (such as the ones mentioned above) using an approach that is mainly and mostly order-theoretic in its essence and nature. I presented the details of this approach in multiple research articles that I posted to arXiv. To summarize the approach though,? recently I presented an abridged outline of the whole approach (as developed so far) in an article available at: http://arxiv.org/abs/1906.11197 I am posting about this outline article to the TYPES mailing list because, first, I'd like to thank all those (mathematicians, PL researchers, language designers, and OO software developers) who gave me their precious feedback on the approach while its development was in its infancy.? That includes thanking in particular David Spivak, Robert Cartwright, John Greiner, Daniel Smith, Larry Paulson, Nathan Carter, and Kariem ElKoush, as well as plenty of the anonymous reviewers of conferences and workshops I submitted some of my work to.? Any shortcomings or errors found in the approach as developed so far or in its presentation should be blamed on me and no one else. And, since the approach has matured a bit, I am now seeking further feedback on the approach from the wider community. Next, given the limited financial and human resources available to me, I am also seeking collaborations with interested and qualified members of the community, which can help in speeding up the development of the approach and in finalizing it sooner. Finally, if you happen to read the outline paper then please make sure that (even if you can provide me with no material support) you do not keep your opinions about the approach to yourself.? All kinds of feedback on the approach --- including constructive, well-justified and respectful criticism --- are more than welcome. Cheers, -Moez PS: Pending finding funding for my trip, I plan to be in Oxford in mid July to attend ACT 2019 and to present a poster there on this work, under the title 'Modeling OO Generics: An Order- and Category-Theoretic Approach'.? Assuming I make it to Oxford, those attending ACT 2019 and interested in a chat should feel free to drop by. From nr at cs.tufts.edu Wed Nov 27 16:38:27 2019 From: nr at cs.tufts.edu (Norman Ramsey) Date: Wed, 27 Nov 2019 16:38:27 -0500 Subject: [TYPES] seeking stupid LaTeX tricks for laying out sets of typing rules Message-ID: <20191127213827.F3643781670@labrador.cs.tufts.edu> Dear Colleagues, I'm looking for advice on preparing figures full of typing rules. What I'd like is to create a figure that has a boxed form of judgment in the top left, then collects all the rules that can prove judgments of the boxed form. At present, I'm using Didier R?my's mathpartir package. The inference rules are nice and readable, and I can collect them easily enough in a `mathpar` environment. But the boxed judgment is placed as if it were just another rule, where it really ought to be in the upper left corner (or some other location which can indicate that it classifies all the rules). I'm sure there must be a trick, but I haven't yet discovered it. How are you typesetting collections of inference rules? Norman From eliasca at kth.se Thu Nov 28 03:07:34 2019 From: eliasca at kth.se (Elias Castegren) Date: Thu, 28 Nov 2019 08:07:34 +0000 Subject: [TYPES] seeking stupid LaTeX tricks for laying out sets of typing rules In-Reply-To: <20191127213827.F3643781670@labrador.cs.tufts.edu> References: <20191127213827.F3643781670@labrador.cs.tufts.edu> Message-ID: Hi! Even though it also has capabilities for generating theorem prover code, I mainly use Ott [1] for writing and generating nice-looking type rules in LaTeX. Together with Jesse Tov?s ottalt package, it gives me rules which are as pretty as I could hope for, and in the style that you?re asking for (I think it is using mathpartir under the hood). Having syntax and grammar checking to make sure that you?re not making silly mistakes is a nice bonus. I still typeset grammars by hand though, as the ones produced by Ott tend to be a bit too whitespace hungry. If you don?t want to introduce an external tool to your workflow, you could look at the code that Ott (with ottalt) generates and borrow the tricks you need. Cheers /Elias [1] https://www.cl.cam.ac.uk/~pes20/ott/ 27 nov. 2019 kl. 22:38 skrev Norman Ramsey >: [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] Dear Colleagues, I'm looking for advice on preparing figures full of typing rules. What I'd like is to create a figure that has a boxed form of judgment in the top left, then collects all the rules that can prove judgments of the boxed form. At present, I'm using Didier R?my's mathpartir package. The inference rules are nice and readable, and I can collect them easily enough in a `mathpar` environment. But the boxed judgment is placed as if it were just another rule, where it really ought to be in the upper left corner (or some other location which can indicate that it classifies all the rules). I'm sure there must be a trick, but I haven't yet discovered it. How are you typesetting collections of inference rules? Norman From gabriel.scherer at gmail.com Thu Nov 28 03:11:13 2019 From: gabriel.scherer at gmail.com (Gabriel Scherer) Date: Thu, 28 Nov 2019 09:11:13 +0100 Subject: [TYPES] seeking stupid LaTeX tricks for laying out sets of typing rules In-Reply-To: <20191127213827.F3643781670@labrador.cs.tufts.edu> References: <20191127213827.F3643781670@labrador.cs.tufts.edu> Message-ID: My approach is to put the judgment-declaration box on its own line, where it ends up centered. With mathpartir ( http://cristal.inria.fr/~remy/latex/mathpartir.html ), a blank line acts as a "normal separation" between elements (which are fit automatically on lines), while \\ forces a break, so I would use \begin{mathpar} \fbox{judgment1}~\text{explanation} \\ rule1 rule2 rule3 \\ \fbox{judgment2}~text{explanation} \\ rule1 rule2 [...] \end{mathpar} On some occasions where I wanted to left-align or right-align an object on a mathpar line, I have used unbreakable spaces as "fake" elements to consume space (note: \and is equivalent ot a newline): \\ left-aligned object ~\and~\and~\and~\and~ right-aligned object \\ It's a very ugly hack, and it works reasonably well. On Thu, Nov 28, 2019 at 6:50 AM Norman Ramsey wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear Colleagues, > > I'm looking for advice on preparing figures full of typing rules. > What I'd like is to create a figure that has a boxed form of judgment > in the top left, then collects all the rules that can prove judgments > of the boxed form. > > At present, I'm using Didier R?my's mathpartir package. > The inference rules are nice and readable, and I can collect them > easily enough in a `mathpar` environment. But the boxed judgment is > placed as if it were just another rule, where it really ought to be in > the upper left corner (or some other location which can indicate that > it classifies all the rules). I'm sure there must be a trick, but I > haven't yet discovered it. > > How are you typesetting collections of inference rules? > > > Norman > > From phil.sieczkowski at gmail.com Thu Nov 28 08:14:41 2019 From: phil.sieczkowski at gmail.com (Filip Sieczkowski) Date: Thu, 28 Nov 2019 14:14:41 +0100 Subject: [TYPES] seeking stupid LaTeX tricks for laying out sets of typing rules In-Reply-To: <20191127213827.F3643781670@labrador.cs.tufts.edu> References: <20191127213827.F3643781670@labrador.cs.tufts.edu> Message-ID: Hi, I usually like to put the judgment form on its own line, similar to what Gabriel suggests, and in that case one solution is to take the judgment form outside the mathpar environment and use something like the flushleft/flushright environments and an \fbox. This works reasonably well unless you define several judgments at the same time, and would like the judgment forms grouped together, since \fbox doesn't do paragraphs. One (somewhat nasty) workaround is to stick the forms in the aligned environment, like so: \begin{flushleft} \fbox{$ \begin{aligned} &\judgmentForm1\\ &\judgmentForm2 %etc. \end{aligned}$} \end{flushleft} \begin{mathpar} %rules go here \end{mathpar} This works reasonably well, although of course if you define a lot of judgments simultaneously it tends to consume a fair bit of space. However, your question got me experimenting, and in the end the clue lies in Didier's documentation for mathpar: it's a normal paragraph with some clever spacing mechanisms. This means that most commands you would expect to work within a paragraph work here too ? including, in particular, \hfill. Thus, the easy approach is to stick the judgment form on the left and leave the rest of the line blank: \begin{mathpar} \fbox{$\judgmentForm$}\hfill\\ %rules \end{mathpar} You can stick some of the rules on the same line, too, but this would require some fiddling with fills, as \hfill is more aggressive than the spacing macros in mathpar (otherwise the trick wouldn't work at all), and the rules will likely get squashed closer together than you would like. It's still a bit of a hack, but at the very least the form with the judgment on its own line should work reliably. Hope this is of some use, ? Filip On Thu, Nov 28, 2019 at 6:47 AM Norman Ramsey wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear Colleagues, > > I'm looking for advice on preparing figures full of typing rules. > What I'd like is to create a figure that has a boxed form of judgment > in the top left, then collects all the rules that can prove judgments > of the boxed form. > > At present, I'm using Didier R?my's mathpartir package. > The inference rules are nice and readable, and I can collect them > easily enough in a `mathpar` environment. But the boxed judgment is > placed as if it were just another rule, where it really ought to be in > the upper left corner (or some other location which can indicate that > it classifies all the rules). I'm sure there must be a trick, but I > haven't yet discovered it. > > How are you typesetting collections of inference rules? > > > Norman > > -- Filip Sieczkowski http://cs.au.dk/~filips e-mail, JID: phil.sieczkowski at gmail.com phone: +48 603289181 "Two roads diverged in a wood, and I-- I chose the one less traveled by, And that has made all the difference." -- Robert Frost, "The road not taken" From wjb at williamjbowman.com Thu Nov 28 13:27:25 2019 From: wjb at williamjbowman.com (William J. Bowman) Date: Thu, 28 Nov 2019 10:27:25 -0800 Subject: [TYPES] seeking stupid LaTeX tricks for laying out sets of typing rules In-Reply-To: References: <20191127213827.F3643781670@labrador.cs.tufts.edu> Message-ID: <20191128182725.GA71392@williamjbowman.com> On Thu, Nov 28, 2019 at 02:14:41PM +0100, Filip Sieczkowski wrote: > \begin{flushleft} > \fbox{$ > \begin{aligned} > &\judgmentForm1\\ > &\judgmentForm2 %etc. > \end{aligned}$} > \end{flushleft} > \begin{mathpar} > %rules go here > \end{mathpar} I use this approach (wrapped in a \newcommand for convenience): \newcommand{\judgshape}[2][]{\begin{flushleft}\fbox{\(#2\)~#1}\end{flushleft}} ... \judgshap[(where \(A\) is well formed)]{\Gamma \vdash e : A} \begin{mathpar} ... \end{mathpar} It works reliably across every format I've submitted with. -- William J. Bowman From Peter.Sewell at cl.cam.ac.uk Thu Nov 28 13:45:35 2019 From: Peter.Sewell at cl.cam.ac.uk (Peter Sewell) Date: Thu, 28 Nov 2019 18:45:35 +0000 Subject: [TYPES] seeking stupid LaTeX tricks for laying out sets of typing rules In-Reply-To: References: <20191127213827.F3643781670@labrador.cs.tufts.edu> Message-ID: The default Ott style using Rok Strnisa and Matt Parkinson's ottlayout.sty (in the Ott distro) also does what you want, as far as I can see - the plain default puts rules one-per-line, while ottlayout lets one choose whether to do that or just lay rules out like words in text (and also lets one control many other things). The \fbox'd judgement is just slapped down as a line or as the first thing, respectively. Peter On Thu, 28 Nov 2019 at 08:12, Elias Castegren wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Hi! > > Even though it also has capabilities for generating theorem prover code, > I mainly use Ott [1] for writing and generating nice-looking type rules > in LaTeX. Together with Jesse Tov?s ottalt package, it gives me rules which > are as pretty as I could hope for, and in the style that you?re asking for > (I think > it is using mathpartir under the hood). Having syntax and grammar checking > to > make sure that you?re not making silly mistakes is a nice bonus. I still > typeset > grammars by hand though, as the ones produced by Ott tend to be a bit too > whitespace hungry. > > If you don?t want to introduce an external tool to your workflow, you > could look > at the code that Ott (with ottalt) generates and borrow the tricks you > need. > > Cheers > > /Elias > > [1] https://www.cl.cam.ac.uk/~pes20/ott/ > > 27 nov. 2019 kl. 22:38 skrev Norman Ramsey nr at cs.tufts.edu>>: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear Colleagues, > > I'm looking for advice on preparing figures full of typing rules. > What I'd like is to create a figure that has a boxed form of judgment > in the top left, then collects all the rules that can prove judgments > of the boxed form. > > At present, I'm using Didier R?my's mathpartir package. > The inference rules are nice and readable, and I can collect them > easily enough in a `mathpar` environment. But the boxed judgment is > placed as if it were just another rule, where it really ought to be in > the upper left corner (or some other location which can indicate that > it classifies all the rules). I'm sure there must be a trick, but I > haven't yet discovered it. > > How are you typesetting collections of inference rules? > > > Norman > > > From nr at cs.tufts.edu Wed Dec 4 10:34:00 2019 From: nr at cs.tufts.edu (Norman Ramsey) Date: Wed, 04 Dec 2019 10:34:00 -0500 Subject: [TYPES] Summary of tricks for setting figures of rules in LaTeX Message-ID: <20191204153401.04FBB900DD7@homedog.cs.tufts.edu> Thanks to the many who replied to my questions about setting figures full of typing rules. Your suggestions were as follows: * \fbox left justified on a line by itself just before \mathpar (possibly with {} \and ... to make room) Example detail: \newcommand{\judgshape}[2][]{\begin{flushleft}\fbox{\(#2\)~#1}\end{flushleft}} ... \judgshap[(where \(A\) is well formed)]{\Gamma \vdash e : A} \begin{mathpar} ... \end{mathpar} * \fbox on its own line in {mathpar} and thus centered * \fbox in mathpar with very ugly hack to force left or right alignment * \fbox in mathpar with nice hack to force left alignment: \fbox{$e \cbvstep e'$}\hfill\\ * Check out back end of `ottalt` package (used with Ott) or ottlayout.sty Thanks again! Norman From gabriel.scherer at gmail.com Fri Dec 20 08:53:05 2019 From: gabriel.scherer at gmail.com (Gabriel Scherer) Date: Fri, 20 Dec 2019 14:53:05 +0100 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles Message-ID: Dear types-list and SIGPLAN, I have long been of the opinion that our scientific publications should be Open Access, and that editors should not request more than a fair price (cost of publication, which Dasgtuhl estimates at $60 per article). In particular, I believe that copyright transfer agreements, as imposed by most editors including the ACM, is deeply unethical: the publishers are not the authors of our scientific production and they should not force us to give our copyright to them. A non-exclusive publishing agreement should be enough. Whether or not you agree with this position, you may be interested in the content of the following letter to the US White House that a coalition of scientific publishers, *including the ACM*, signed and support. https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf press release from the coalition of editors: https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles (This letter was written in the context of a proposed US legislation to force more scientists to publish their work in Fair Open Access venues. I haven't been able to find a precise link to a discussion of this proposed legislation.) The following parts of the letter co-signed by the ACM are particularly juicy: > [We] have learned that the Administration may be preparing to step > into the private marketplace and force the immediate free distribution > of journal articles financed and published by organizations in the > private sector, including many non-profits. This would effectively > nationalize the valuable American intellectual property that we > produce and force us to give it away to the rest of the world for > free. > This mandate [...] would make it very difficult for most American > publishers to invest in publishing these articles. As a consequence, > it would place increased financial responsibility on the government > through diverted federal research grant funds or additional monies > to underwrite the important value added by publishing. In the coming > years, this cost shift would place billions of dollars of new and > additional burden on taxpayers. In my discussion with many of us, I regularly hear that the ACM is "not evil" (the SIGPLAN, of course, is pure good!) and that placating its weird views (for example, that it really does cost $700 or $900 to publish an article as Open Source) is good for our research community. It do not see how this argument is compatible with the ACM signing this letter. I believe that many of our activities, which we collectively trained ourselves to see as harmless administrative details of our research work, are in fact empowering the ACM to make those claims. Should we accept to give away our copyright, or pay? unreasonable Gold Access author processing charges (APCs)? ?: The SIGPLAN decision to cover APC costs for PACMPL articles is shielding many of us from paying APCs. But many of the smaller conferences, symposiums or workshops in our community whose proceedings are handled by the ACM are still limited to "pay $900" (or "pay $25 per page") as the only Open Access option, with copyright transfer as the only free choice, which is effectively keeping those proceedings Closed-Access. From jonathan.aldrich at cs.cmu.edu Fri Dec 20 09:31:48 2019 From: jonathan.aldrich at cs.cmu.edu (Jonathan Aldrich) Date: Fri, 20 Dec 2019 09:31:48 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: Message-ID: I agree with Gabriel. Furthermore, I think we should do something. Specifically, I think it would be great if we (SIGPLAN/members of types-list) could coordinate an open letter--which might eventually include other allied communities within ACM--expressing how problematic it is for ACM to sign this statement, and how it deeply misrepresents the will of ACM membership. We should ask for ACM to (1) disavow the Publishers' statement, and for them to (2) set open access fees at a rate that reflects the cost of maintaining and making the works indefinitely available--but at a rate that does *not* subsidize other "good works" of ACM (that is the reason for the current high price; the ACM digital library has subsidized other good but unrelated things for a long time, and ACM sets the price of open access fees to match the overall "profitability" of the library. I strongly support many of those other good things, e.g. helping students attend conferences, but I do not think "taxing" open access is an appropriate way to fund them). I think we should not take the most extreme stances--for example, I personally agree with Garbriel that there is an ethical problem with closed access publication models, but I do not think everyone would go that far, and I think we would be smarter to focus a letter on things that would get very wide agreement at least within the SIGPLAN/Types community, and hopefully within ACM at large. We might then get many signatures, perhaps enough to send a strong message. ACM leadership does, after all, work for its membership--us! I'd be willing to help draft such a thing, but it would be best to have SIGPLAN leadership and/or ACM Fellow/Turing Award people on board. Is there anyone in that category who could lend support or even take the lead on something like this? Feel free to respond privately to me (in which case I will later post a summary) and/or publicly to the list with further suggestions. Best, Jonathan On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > Dear types-list and SIGPLAN, > > I have long been of the opinion that our scientific publications > should be Open Access, and that editors should not request more than > a fair price (cost of publication, which Dasgtuhl estimates at $60 > per article). In particular, I believe that copyright transfer > agreements, as imposed by most editors including the ACM, is deeply > unethical: the publishers are not the authors of our scientific > production and they should not force us to give our copyright to > them. A non-exclusive publishing agreement should be enough. > > Whether or not you agree with this position, you may be interested in > the content of the following letter to the US White House that > a coalition of scientific publishers, *including the ACM*, signed and > support. > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > press release from the coalition of editors: > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > (This letter was written in the context of a proposed US legislation > to force more scientists to publish their work in Fair Open Access > venues. I haven't been able to find a precise link to a discussion of > this proposed legislation.) > > The following parts of the letter co-signed by the ACM are > particularly juicy: > > > [We] have learned that the Administration may be preparing to step > > into the private marketplace and force the immediate free distribution > > of journal articles financed and published by organizations in the > > private sector, including many non-profits. This would effectively > > nationalize the valuable American intellectual property that we > > produce and force us to give it away to the rest of the world for > > free. > > > This mandate [...] would make it very difficult for most American > > publishers to invest in publishing these articles. As a consequence, > > it would place increased financial responsibility on the government > > through diverted federal research grant funds or additional monies > > to underwrite the important value added by publishing. In the coming > > years, this cost shift would place billions of dollars of new and > > additional burden on taxpayers. > > In my discussion with many of us, I regularly hear that the ACM is > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > its weird views (for example, that it really does cost $700 or $900 to > publish an article as Open Source) is good for our research > community. It do not see how this argument is compatible with the ACM > signing this letter. > > I believe that many of our activities, which we collectively trained > ourselves to see as harmless administrative details of our research > work, are in fact empowering the ACM to make those claims. Should we > accept to give away our copyright, or pay? unreasonable > Gold Access author processing charges (APCs)? > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > shielding many of us from paying APCs. But many of the smaller > conferences, symposiums or workshops in our community whose > proceedings are handled by the ACM are still limited to "pay > $900" (or "pay $25 per page") as the only Open Access option, with > copyright transfer as the only free choice, which is effectively > keeping those proceedings Closed-Access. > From tim.sweeney at epicgames.com Fri Dec 20 09:46:46 2019 From: tim.sweeney at epicgames.com (Tim Sweeney) Date: Fri, 20 Dec 2019 09:46:46 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: Message-ID: Thanks for standing up for this. Epic Games (home to few type theorists but many ACM SigGraph participants) will make our support for authors' right to freely distribute their work known to ACM. -Tim On Fri, Dec 20, 2019 at 9:40 AM Jonathan Aldrich < jonathan.aldrich at cs.cmu.edu> wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > I agree with Gabriel. Furthermore, I think we should do something. > > Specifically, I think it would be great if we (SIGPLAN/members of > types-list) could coordinate an open letter--which might eventually include > other allied communities within ACM--expressing how problematic it is for > ACM to sign this statement, and how it deeply misrepresents the will of ACM > membership. We should ask for ACM to (1) disavow the Publishers' > statement, and for them to (2) set open access fees at a rate that reflects > the cost of maintaining and making the works indefinitely available--but at > a rate that does *not* subsidize other "good works" of ACM (that is the > reason for the current high price; the ACM digital library has subsidized > other good but unrelated things for a long time, and ACM sets the price of > open access fees to match the overall "profitability" of the library. I > strongly support many of those other good things, e.g. helping students > attend conferences, but I do not think "taxing" open access is an > appropriate way to fund them). > > I think we should not take the most extreme stances--for example, I > personally agree with Garbriel that there is an ethical problem with closed > access publication models, but I do not think everyone would go that far, > and I think we would be smarter to focus a letter on things that would get > very wide agreement at least within the SIGPLAN/Types community, and > hopefully within ACM at large. We might then get many signatures, perhaps > enough to send a strong message. ACM leadership does, after all, work for > its membership--us! > > I'd be willing to help draft such a thing, but it would be best to have > SIGPLAN leadership and/or ACM Fellow/Turing Award people on board. Is > there anyone in that category who could lend support or even take the lead > on something like this? Feel free to respond privately to me (in which > case I will later post a summary) and/or publicly to the list with further > suggestions. > > Best, > > Jonathan > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer > > wrote: > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > ] > > > > Dear types-list and SIGPLAN, > > > > I have long been of the opinion that our scientific publications > > should be Open Access, and that editors should not request more than > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > per article). In particular, I believe that copyright transfer > > agreements, as imposed by most editors including the ACM, is deeply > > unethical: the publishers are not the authors of our scientific > > production and they should not force us to give our copyright to > > them. A non-exclusive publishing agreement should be enough. > > > > Whether or not you agree with this position, you may be interested in > > the content of the following letter to the US White House that > > a coalition of scientific publishers, *including the ACM*, signed and > > support. > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > press release from the coalition of editors: > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > (This letter was written in the context of a proposed US legislation > > to force more scientists to publish their work in Fair Open Access > > venues. I haven't been able to find a precise link to a discussion of > > this proposed legislation.) > > > > The following parts of the letter co-signed by the ACM are > > particularly juicy: > > > > > [We] have learned that the Administration may be preparing to step > > > into the private marketplace and force the immediate free distribution > > > of journal articles financed and published by organizations in the > > > private sector, including many non-profits. This would effectively > > > nationalize the valuable American intellectual property that we > > > produce and force us to give it away to the rest of the world for > > > free. > > > > > This mandate [...] would make it very difficult for most American > > > publishers to invest in publishing these articles. As a consequence, > > > it would place increased financial responsibility on the government > > > through diverted federal research grant funds or additional monies > > > to underwrite the important value added by publishing. In the coming > > > years, this cost shift would place billions of dollars of new and > > > additional burden on taxpayers. > > > > In my discussion with many of us, I regularly hear that the ACM is > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > its weird views (for example, that it really does cost $700 or $900 to > > publish an article as Open Source) is good for our research > > community. It do not see how this argument is compatible with the ACM > > signing this letter. > > > > I believe that many of our activities, which we collectively trained > > ourselves to see as harmless administrative details of our research > > work, are in fact empowering the ACM to make those claims. Should we > > accept to give away our copyright, or pay? unreasonable > > Gold Access author processing charges (APCs)? > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > shielding many of us from paying APCs. But many of the smaller > > conferences, symposiums or workshops in our community whose > > proceedings are handled by the ACM are still limited to "pay > > $900" (or "pay $25 per page") as the only Open Access option, with > > copyright transfer as the only free choice, which is effectively > > keeping those proceedings Closed-Access. > > > From monnier at iro.umontreal.ca Fri Dec 20 10:40:31 2019 From: monnier at iro.umontreal.ca (Stefan Monnier) Date: Fri, 20 Dec 2019 10:40:31 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: (Gabriel Scherer's message of "Fri, 20 Dec 2019 14:53:05 +0100") References: Message-ID: > Whether or not you agree with this position, you may be interested in > the content of the following letter to the US White House that > a coalition of scientific publishers, *including the ACM*, signed and > support. Thanks Gabriel. I find this letter unacceptable, indeed. I sent a complaint to acmhelp at acm.org about it, not sure if it's the best place. Stefan From nikhil at acm.org Fri Dec 20 12:03:02 2019 From: nikhil at acm.org (Rishiyur Nikhil) Date: Fri, 20 Dec 2019 12:03:02 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: Message-ID: > I sent a complaint to acmhelp at acm.org about it, not sure if it's the best place. You can also follow and participate in a conversation on this at the ACM President's Twitter feed, at: https://twitter.com/ACM_president Nikhil From roberto at dicosmo.org Fri Dec 20 13:08:07 2019 From: roberto at dicosmo.org (Roberto Di Cosmo) Date: Fri, 20 Dec 2019 19:08:07 +0100 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: Message-ID: <20191220180807.GA9697@traveler> Thanks Gabriel for bringing this to this list: it was indeed shocking to see ACM (and many other learned societies) in the list of signatories of this letter. The fact that many small learned societies do not feel ready to jump into a pure open access model right away does not justify their signature on a letter containing highly debatable (that's an euphemism) statements like the ones you pinpoint. By a curious coincidence, I got almost at the same time an ACM newlsetter (Blue Diamond) containing among other announcements, this one: ACM OPEN: A New Transformative Model for Open Access Publication Over the past year ACM Publications staff have been working collaboratively with a group of large research universities in the United States to develop an entirely new and innovative model for Open Access publication that has the potential to transition ACM into a predominantly Open Access publisher over the next decade or sooner. You can find details of the proposed model at https://www.acm.org/publications/openaccess#acmopen Cheers -- Roberto On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear types-list and SIGPLAN, > > I have long been of the opinion that our scientific publications > should be Open Access, and that editors should not request more than > a fair price (cost of publication, which Dasgtuhl estimates at $60 > per article). In particular, I believe that copyright transfer > agreements, as imposed by most editors including the ACM, is deeply > unethical: the publishers are not the authors of our scientific > production and they should not force us to give our copyright to > them. A non-exclusive publishing agreement should be enough. > > Whether or not you agree with this position, you may be interested in > the content of the following letter to the US White House that > a coalition of scientific publishers, *including the ACM*, signed and > support. > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > press release from the coalition of editors: > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > (This letter was written in the context of a proposed US legislation > to force more scientists to publish their work in Fair Open Access > venues. I haven't been able to find a precise link to a discussion of > this proposed legislation.) > > The following parts of the letter co-signed by the ACM are > particularly juicy: > > > [We] have learned that the Administration may be preparing to step > > into the private marketplace and force the immediate free distribution > > of journal articles financed and published by organizations in the > > private sector, including many non-profits. This would effectively > > nationalize the valuable American intellectual property that we > > produce and force us to give it away to the rest of the world for > > free. > > > This mandate [...] would make it very difficult for most American > > publishers to invest in publishing these articles. As a consequence, > > it would place increased financial responsibility on the government > > through diverted federal research grant funds or additional monies > > to underwrite the important value added by publishing. In the coming > > years, this cost shift would place billions of dollars of new and > > additional burden on taxpayers. > > In my discussion with many of us, I regularly hear that the ACM is > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > its weird views (for example, that it really does cost $700 or $900 to > publish an article as Open Source) is good for our research > community. It do not see how this argument is compatible with the ACM > signing this letter. > > I believe that many of our activities, which we collectively trained > ourselves to see as harmless administrative details of our research > work, are in fact empowering the ACM to make those claims. Should we > accept to give away our copyright, or pay? unreasonable > Gold Access author processing charges (APCs)? > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > shielding many of us from paying APCs. But many of the smaller > conferences, symposiums or workshops in our community whose > proceedings are handled by the ACM are still limited to "pay > $900" (or "pay $25 per page") as the only Open Access option, with > copyright transfer as the only free choice, which is effectively > keeping those proceedings Closed-Access. -- Roberto Di Cosmo ------------------------------------------------------------------ Computer Science Professor (on leave at INRIA from IRIF/University Paris Diderot) Director Software Heritage https://www.softwareheritage.org INRIA Bureau C328 E-mail : roberto at dicosmo.org 2, Rue Simone Iff Web page : http://www.dicosmo.org CS 42112 Twitter : http://twitter.com/rdicosmo 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 ------------------------------------------------------------------ GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 From gabriel.scherer at gmail.com Sat Dec 21 06:01:00 2019 From: gabriel.scherer at gmail.com (Gabriel Scherer) Date: Sat, 21 Dec 2019 12:01:00 +0100 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: <20191220180807.GA9697@traveler> References: <20191220180807.GA9697@traveler> Message-ID: Dear Roberto (and list), The new ACM Open model is based on the core idea of saving the licensing revenue of the ACM by shifting costs from their many customers (including in particular companies) to only the institutions who submit the articles. They hope that the academic actors that produce the scientific value will also pay for current ACM expenses. This model is completely incompatible with having fair Open Access prices for ACM publications; on the contrary, it would result in a strong total-cost increase for academic entities that publish in ACM proceedings. This is frankly explained on the (current version of) the ACM Open documentation page: https://www.acm.org/publications/openaccess#acmopen > Today, ACM Publications and the ACM Digital Library platform are funded by > selling "read" or "access" licenses to approximately 2,700 universities, > government research labs, and corporations from around the world. The > income generated from the sale of these licenses [...] is approximately > $20M+ annually > The vast majority of [ACM] articles are authored by individuals affiliated > with ~1,000 institutions, which is roughly 1/3 of the institutions that > license ?access? to the ACM Digital Library. So, the main challenge for ACM > is how to generate roughly the same income from 1/3 the number of > institutions over the long term, as ACM transitions from selling > institutional "access" to an institutional "OA publication" model and more > and more of the articles published in the ACM DL are published in front of > the subscription paywall. > A transition to fair Open Access practices would require the difficult decision of giving up on licensing revenue. The ACM does not seem willing to do it, and cannot be trusted to do it eventually. On Fri, Dec 20, 2019 at 7:08 PM Roberto Di Cosmo wrote: > Thanks Gabriel for bringing this to this list: it was indeed shocking to > see ACM > (and many other learned societies) in the list of signatories of this > letter. > > The fact that many small learned societies do not feel ready to jump into > a pure > open access model right away does not justify their signature on a letter > containing highly debatable (that's an euphemism) statements like the ones > you pinpoint. > > By a curious coincidence, I got almost at the same time an ACM newlsetter > (Blue > Diamond) containing among other announcements, this one: > > ACM OPEN: A New Transformative Model for Open Access Publication > > Over the past year ACM Publications staff have been working > collaboratively with > a group of large research universities in the United States to > develop an > entirely new and innovative model for Open Access publication that > has the > potential to transition ACM into a predominantly Open Access > publisher over the > next decade or sooner. > > You can find details of the proposed model at > https://www.acm.org/publications/openaccess#acmopen > > Cheers > > -- > Roberto > > On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > Dear types-list and SIGPLAN, > > > > I have long been of the opinion that our scientific publications > > should be Open Access, and that editors should not request more than > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > per article). In particular, I believe that copyright transfer > > agreements, as imposed by most editors including the ACM, is deeply > > unethical: the publishers are not the authors of our scientific > > production and they should not force us to give our copyright to > > them. A non-exclusive publishing agreement should be enough. > > > > Whether or not you agree with this position, you may be interested in > > the content of the following letter to the US White House that > > a coalition of scientific publishers, *including the ACM*, signed and > > support. > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > press release from the coalition of editors: > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > (This letter was written in the context of a proposed US legislation > > to force more scientists to publish their work in Fair Open Access > > venues. I haven't been able to find a precise link to a discussion of > > this proposed legislation.) > > > > The following parts of the letter co-signed by the ACM are > > particularly juicy: > > > > > [We] have learned that the Administration may be preparing to step > > > into the private marketplace and force the immediate free distribution > > > of journal articles financed and published by organizations in the > > > private sector, including many non-profits. This would effectively > > > nationalize the valuable American intellectual property that we > > > produce and force us to give it away to the rest of the world for > > > free. > > > > > This mandate [...] would make it very difficult for most American > > > publishers to invest in publishing these articles. As a consequence, > > > it would place increased financial responsibility on the government > > > through diverted federal research grant funds or additional monies > > > to underwrite the important value added by publishing. In the coming > > > years, this cost shift would place billions of dollars of new and > > > additional burden on taxpayers. > > > > In my discussion with many of us, I regularly hear that the ACM is > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > its weird views (for example, that it really does cost $700 or $900 to > > publish an article as Open Source) is good for our research > > community. It do not see how this argument is compatible with the ACM > > signing this letter. > > > > I believe that many of our activities, which we collectively trained > > ourselves to see as harmless administrative details of our research > > work, are in fact empowering the ACM to make those claims. Should we > > accept to give away our copyright, or pay? unreasonable > > Gold Access author processing charges (APCs)? > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > shielding many of us from paying APCs. But many of the smaller > > conferences, symposiums or workshops in our community whose > > proceedings are handled by the ACM are still limited to "pay > > $900" (or "pay $25 per page") as the only Open Access option, with > > copyright transfer as the only free choice, which is effectively > > keeping those proceedings Closed-Access. > > -- > Roberto Di Cosmo > > ------------------------------------------------------------------ > Computer Science Professor > (on leave at INRIA from IRIF/University Paris Diderot) > > Director > Software Heritage https://www.softwareheritage.org > INRIA > Bureau C328 E-mail : roberto at dicosmo.org > 2, Rue Simone Iff Web page : http://www.dicosmo.org > CS 42112 Twitter : http://twitter.com/rdicosmo > 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 > ------------------------------------------------------------------ > > GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 > > From selinger at mathstat.dal.ca Sun Dec 22 01:37:53 2019 From: selinger at mathstat.dal.ca (selinger at mathstat.dal.ca) Date: Sun, 22 Dec 2019 06:37:53 +0000 Subject: [TYPES] In a letter to the US White House, In-Reply-To: Message-ID: I was recently invited to the editorial board of a new ACM journal. I refused to become an editor, on the grounds that the journal is not open access. I feel that as editors, we volunteer our time, and I have decided to do so only for journals that are truly open access. And by "open access", I do not mean some "gold" model based on author charges (currently $1300-$1700 per ACM journal article, see https://www.acm.org/publications/openaccess). I mean journals that are free for both authors and readers. There are already many such journals, and they are usually community-run. Examples include Logical Methods in Computer Science (https://lmcs.episciences.org/), Compositionality (https://www.compositionality-journal.org/), and Quantum (https://quantum-journal.org/). There is relatively little in the way of interested groups of individuals starting new such journals. I think the best way to get rid of unreasonable publishing models is not to participate in them. -- Peter Jonathan Aldrich wrote: > > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > I agree with Gabriel. Furthermore, I think we should do something. > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer > wrote: > > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > > ] > > > > Dear types-list and SIGPLAN, > > > > I have long been of the opinion that our scientific publications > > should be Open Access, and that editors should not request more than > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > per article). In particular, I believe that copyright transfer > > agreements, as imposed by most editors including the ACM, is deeply > > unethical: the publishers are not the authors of our scientific > > production and they should not force us to give our copyright to > > them. A non-exclusive publishing agreement should be enough. > > > > Whether or not you agree with this position, you may be interested in > > the content of the following letter to the US White House that > > a coalition of scientific publishers, *including the ACM*, signed and > > support. > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > press release from the coalition of editors: > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > (This letter was written in the context of a proposed US legislation > > to force more scientists to publish their work in Fair Open Access > > venues. I haven't been able to find a precise link to a discussion of > > this proposed legislation.) > > > > The following parts of the letter co-signed by the ACM are > > particularly juicy: > > > > > [We] have learned that the Administration may be preparing to step > > > into the private marketplace and force the immediate free distribution > > > of journal articles financed and published by organizations in the > > > private sector, including many non-profits. This would effectively > > > nationalize the valuable American intellectual property that we > > > produce and force us to give it away to the rest of the world for > > > free. > > > > > This mandate [...] would make it very difficult for most American > > > publishers to invest in publishing these articles. As a consequence, > > > it would place increased financial responsibility on the government > > > through diverted federal research grant funds or additional monies > > > to underwrite the important value added by publishing. In the coming > > > years, this cost shift would place billions of dollars of new and > > > additional burden on taxpayers. > > > > In my discussion with many of us, I regularly hear that the ACM is > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > its weird views (for example, that it really does cost $700 or $900 to > > publish an article as Open Source) is good for our research > > community. It do not see how this argument is compatible with the ACM > > signing this letter. > > > > I believe that many of our activities, which we collectively trained > > ourselves to see as harmless administrative details of our research > > work, are in fact empowering the ACM to make those claims. Should we > > accept to give away our copyright, or pay? unreasonable > > Gold Access author processing charges (APCs)? > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > shielding many of us from paying APCs. But many of the smaller > > conferences, symposiums or workshops in our community whose > > proceedings are handled by the ACM are still limited to "pay > > $900" (or "pay $25 per page") as the only Open Access option, with > > copyright transfer as the only free choice, which is effectively > > keeping those proceedings Closed-Access. > > > > From andru at cs.cornell.edu Sun Dec 22 09:46:15 2019 From: andru at cs.cornell.edu (Andrew Myers) Date: Sun, 22 Dec 2019 09:46:15 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: <20191220180807.GA9697@traveler> Message-ID: It feels a bit facile to bash the ACM for signing onto this letter. The letter does not mean that they oppose making publications freely available; in fact, I believe open access is a goal for ACM. The letter means that they oppose having the government *mandate* that all scientific publishers operate in this way. Exactly what the right funding model is for scientific publications is still up in the air. Should the government spend taxes enforcing rules whose implications we do?not?fully?understand??I?think?not. The discussions I have seen about this topic seem to focus on the costs to readers and authors while completely ignoring the economics of publishing. I hope we can agree that publishers do provide some value in supporting the scientific process, for example by maintaining archives of publications for decades and across formats. That value can only be delivered if ACM et al. have money. Where are they supposed to get it? The old model of libraries paying ACM subscriptions is dying and is incompatible with open access. Corporate charity is unreliable and insufficient. The only other player with an incentive to provide money is the authors. My understanding is that the economics are forcing ACM to?go?in?that?direction. I believe ACM Is trying to be a good actor here, unlike publishers that double-dip by extracting money from both the authors (publication fees) and the readers (subscription fees); those publishers are doing very well financially and generating well-earned resentment. My understanding is that ACM does not want to double-dip. Instead, the idea is that authors at institutions with ACM subscriptions will pay lower or no fees for publications. That should keep the total cost to institutions under control and hopefully approximately cost-neutral. And note that the open access fees charged to other authors are still much lower than the author fees charged by other publishers. The journal Nature charges authors?$2000,?for?example,?and?it?is?not?the?high?end. Best, Andrew?Myers Gabriel?Scherer?wrote?on?12/21/19?6:01?AM: > [?The?Types?Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list?] > > Dear?Roberto?(and?list), > > The?new?ACM?Open?model?is?based?on?the?core?idea?of?saving?the?licensing > revenue?of?the?ACM?by?shifting?costs?from?their?many?customers?(including > in?particular?companies)?to?only?the?institutions?who?submit?the?articles. > > They?hope?that?the?academic?actors?that?produce?the?scientific?value?will > also?pay?for?current?ACM?expenses.?This?model?is?completely?incompatible > with?having?fair?Open?Access?prices?for?ACM?publications;?on?the?contrary, > > it?would?result?in?a?strong?total-cost?increase?for?academic?entities?that > > publish?in?ACM?proceedings. > > This?is?frankly?explained?on?the?(current?version?of)?the?ACM?Open > documentation?page: > https://www.acm.org/publications/openaccess#acmopen > >> Today,?ACM?Publications?and?the?ACM?Digital?Library?platform?are?funded?by >> >> selling?"read"?or?"access"?licenses?to?approximately?2,700?universities, >> government?research?labs,?and?corporations?from?around?the?world.?The >> income?generated?from?the?sale?of?these?licenses?[...]?is?approximately >> $20M+?annually >> > The?vast?majority?of?[ACM]?articles?are?authored?by?individuals?affiliated > >> with?~1,000?institutions,?which?is?roughly?1/3?of?the?institutions?that >> license??access??to?the?ACM?Digital?Library.?So,?the?main?challenge?for?ACM >> >> is?how?to?generate?roughly?the?same?income?from?1/3?the?number?of >> institutions?over?the?long?term,?as?ACM?transitions?from?selling >> institutional?"access"?to?an?institutional?"OA?publication"?model?and?more >> >> and?more?of?the?articles?published?in?the?ACM?DL?are?published?in?front?of >> >> the?subscription?paywall. >> > A?transition?to?fair?Open?Access?practices?would?require?the?difficult > decision?of?giving?up?on?licensing?revenue. > The?ACM?does?not?seem?willing?to?do?it,?and?cannot?be?trusted?to?do?it > eventually. > > > On?Fri,?Dec?20,?2019?at?7:08?PM?Roberto?Di?Cosmo > wrote: > >> Thanks?Gabriel?for?bringing?this?to?this?list:?it?was?indeed?shocking?to >> see?ACM >> (and?many?other?learned?societies)?in?the?list?of?signatories?of?this >> letter. >> >> The?fact?that?many?small?learned?societies?do?not?feel?ready?to?jump?into >> >> a?pure >> open?access?model?right?away?does?not?justify?their?signature?on?a?letter >> >> containing?highly?debatable?(that's?an?euphemism)?statements?like?the?ones >> >> you?pinpoint. >> >> By?a?curious?coincidence,?I?got?almost?at?the?same?time?an?ACM?newlsetter >> >> (Blue >> Diamond)?containing?among?other?announcements,?this?one: >> >> ?????ACM?OPEN:?A?New?Transformative?Model?for?Open?Access?Publication >> >> ??????Over?the?past?year?ACM?Publications?staff?have?been?working >> collaboratively?with >> ??????a?group?of?large?research?universities?in?the?United?States?to >> develop?an >> ??????entirely?new?and?innovative?model?for?Open?Access?publication?that >> has?the >> ??????potential?to?transition?ACM?into?a?predominantly?Open?Access >> publisher?over?the >> ??????next?decade?or?sooner. >> >> You?can?find?details?of?the?proposed?model?at >> https://www.acm.org/publications/openaccess#acmopen >> >> Cheers >> >> -- >> Roberto >> >> On?Fri,?Dec?20,?2019?at?02:53:05PM?+0100,?Gabriel?Scherer?wrote: >>> [?The?Types?Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>> Dear?types-list?and?SIGPLAN, >>> >>> I?have?long?been?of?the?opinion?that?our?scientific?publications >>> should?be?Open?Access,?and?that?editors?should?not?request?more?than >>> a?fair?price?(cost?of?publication,?which?Dasgtuhl?estimates?at?$60 >>> per?article).?In?particular,?I?believe?that?copyright?transfer >>> agreements,?as?imposed?by?most?editors?including?the?ACM,?is?deeply >>> unethical:?the?publishers?are?not?the?authors?of?our?scientific >>> production?and?they?should?not?force?us?to?give?our?copyright?to >>> them.?A?non-exclusive?publishing?agreement?should?be?enough. >>> >>> Whether?or?not?you?agree?with?this?position,?you?may?be?interested?in >>> the?content?of?the?following?letter?to?the?US?White?House?that >>> a?coalition?of?scientific?publishers,?*including?the?ACM*,?signed?and >>> support. >>> >>> >>> >> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >> >>> ???press?release?from?the?coalition?of?editors: >>> >> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >> >>> (This?letter?was?written?in?the?context?of?a?proposed?US?legislation >>> to?force?more?scientists?to?publish?their?work?in?Fair?Open?Access >>> venues.?I?haven't?been?able?to?find?a?precise?link?to?a?discussion?of >>> this?proposed?legislation.) >>> >>> The?following?parts?of?the?letter?co-signed?by?the?ACM?are >>> particularly?juicy: >>> >>>> [We]?have?learned?that?the?Administration?may?be?preparing?to?step >>>> into?the?private?marketplace?and?force?the?immediate?free?distribution >>>> of?journal?articles?financed?and?published?by?organizations?in?the >>>> private?sector,?including?many?non-profits.?This?would?effectively >>>> nationalize?the?valuable?American?intellectual?property?that?we >>>> produce?and?force?us?to?give?it?away?to?the?rest?of?the?world?for >>>> free. >>>> This?mandate?[...]?would?make?it?very?difficult?for?most?American >>>> publishers?to?invest?in?publishing?these?articles.?As?a?consequence, >>>> it?would?place?increased?financial?responsibility?on?the?government >>>> through?diverted?federal?research?grant?funds?or?additional?monies >>>> to?underwrite?the?important?value?added?by?publishing.?In?the?coming >>>> years,?this?cost?shift?would?place?billions?of?dollars?of?new?and >>>> additional?burden?on?taxpayers. >>> In?my?discussion?with?many?of?us,?I?regularly?hear?that?the?ACM?is >>> "not?evil"?(the?SIGPLAN,?of?course,?is?pure?good!)?and?that?placating >>> its?weird?views?(for?example,?that?it?really?does?cost?$700?or?$900?to >>> publish?an?article?as?Open?Source)?is?good?for?our?research >>> community.?It?do?not?see?how?this?argument?is?compatible?with?the?ACM >>> signing?this?letter. >>> >>> I?believe?that?many?of?our?activities,?which?we?collectively?trained >>> ourselves?to?see?as?harmless?administrative?details?of?our?research >>> work,?are?in?fact?empowering?the?ACM?to?make?those?claims.?Should?we >>> accept?to?give?away?our?copyright,?or?pay??unreasonable >>> Gold?Access?author?processing?charges?(APCs)? >>> >>> ?:?The?SIGPLAN?decision?to?cover?APC?costs?for?PACMPL?articles?is >>> shielding?many?of?us?from?paying?APCs.?But?many?of?the?smaller >>> conferences,?symposiums?or?workshops?in?our?community?whose >>> proceedings?are?handled?by?the?ACM?are?still?limited?to?"pay >>> $900"?(or?"pay?$25?per?page")?as?the?only?Open?Access?option,?with >>> copyright?transfer?as?the?only?free?choice,?which?is?effectively >>> keeping?those?proceedings?Closed-Access. >> -- >> Roberto?Di?Cosmo >> >> ------------------------------------------------------------------ >> Computer?Science?Professor >> ?????????????(on?leave?at?INRIA?from?IRIF/University?Paris?Diderot) >> >> Director >> Software?Heritage https://www.softwareheritage.org >> INRIA >> Bureau?C328??????????????????E-mail?: roberto at dicosmo.org >> 2,?Rue?Simone?Iff??????????Web?page?: http://www.dicosmo.org >> CS?42112????????????????????Twitter?: http://twitter.com/rdicosmo >> 75589?Paris?Cedex?12????????????Tel?:?+33?1?80?49?44?42 >> ------------------------------------------------------------------ >> >> GPG?fingerprint?2931?20CE?3A5A?5390?98EC?8BFC?FCCA?C3BE?39CB?12D3 >> >> From gabriel.scherer at gmail.com Sun Dec 22 12:52:12 2019 From: gabriel.scherer at gmail.com (Gabriel Scherer) Date: Sun, 22 Dec 2019 18:52:12 +0100 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: <20191220180807.GA9697@traveler> Message-ID: Dear Andrew (and list), I believe open access is a goal for ACM This is what the ACM says, but this is not their actions suggest. Some examples: 1. They signed this letter. (They defend their choice in https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) 2. Events affiliated with an ACM conference, such as a workshop, are not allowed to publish their proceedings as (fair) open-access if they wish to, for example by publishing in ETCS or LiPICS. (I know from my experience with the ML and OCaml workshops that ACM people check this and enforce this rule.) 3. According to private communication with ETAPS organizers, the Gold Open Access deal offered by Springer costs *less* per paper for ETAPS than the Open Access model that SIGPLAN generously funds for PACMPL. If you're doing worse than Springer at Open Access, you are probably not trying very hard. I hope we can agree that publishers do provide some value in supporting the > scientific process, for example by maintaining archives of publications for > decades and across formats. > According to LiPICS (the fair Open Access publishing arm of Schloss Dagstuhl), their edition/typesetting work costs 60? per article ( https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). (In any case, ACM outsources their edition work on proceedings to external companies, that if I understand correctly are budgeted as part of the conference organization, so not paid by ACM itself.) According to arXiv, their long-term archival platform costs <$7 per article ( https://arxiv.org/help/support/whitepaper#21-budget ). On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers wrote: > It feels a bit facile to bash the ACM for signing onto this letter. The > letter does not mean that they oppose making publications freely available; > in fact, I believe open access is a goal for ACM. The letter means that > they oppose having the government *mandate* that all scientific publishers > operate in this way. Exactly what the right funding model is for scientific > publications is still up in the air. Should the government spend taxes > enforcing rules whose implications we do not fully understand? I think not. > > The discussions I have seen about this topic seem to focus on the costs to > readers and authors while completely ignoring the economics of publishing. > I hope we can agree that publishers do provide some value in supporting the > scientific process, for example by maintaining archives of publications for > decades and across formats. That value can only be delivered if ACM et al. > have money. Where are they supposed to get it? The old model of libraries > paying ACM subscriptions is dying and is incompatible with open access. > Corporate charity is unreliable and insufficient. The only other player > with an incentive to provide money is the authors. My understanding is that > the economics are forcing ACM to go in that direction. > > I believe ACM Is trying to be a good actor here, unlike publishers that > double-dip by extracting money from both the authors (publication fees) and > the readers (subscription fees); those publishers are doing very well > financially and generating well-earned resentment. My understanding is that > ACM does not want to double-dip. Instead, the idea is that authors at > institutions with ACM subscriptions will pay lower or no fees for > publications. That should keep the total cost to institutions under control > and hopefully approximately cost-neutral. And note that the open access > fees charged to other authors are still much lower than the author fees > charged by other publishers. The journal Nature charges > authors $2000, for example, and it is not the high end. > > Best, > > Andrew Myers > > Gabriel Scherer wrote on 12/21/19 6:01 AM: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > Dear Roberto (and list), > > The new ACM Open model is based on the core idea of saving the licensing > revenue of the ACM by shifting costs from their many customers (including > in particular companies) to only the institutions who submit the articles. > They hope that the academic actors that produce the scientific value will > also pay for current ACM expenses. This model is completely incompatible > with having fair Open Access prices for ACM publications; on the contrary, > it would result in a strong total-cost increase for academic entities that > publish in ACM proceedings. > > This is frankly explained on the (current version of) the ACM Open > documentation page: > https://www.acm.org/publications/openaccess#acmopen > > Today, ACM Publications and the ACM Digital Library platform are funded by > selling "read" or "access" licenses to approximately 2,700 universities, > government research labs, and corporations from around the world. The > income generated from the sale of these licenses [...] is approximately > $20M+ annually > > The vast majority of [ACM] articles are authored by individuals affiliated > > with ~1,000 institutions, which is roughly 1/3 of the institutions that > license ?access? to the ACM Digital Library. So, the main challenge for ACM > > is how to generate roughly the same income from 1/3 the number of > institutions over the long term, as ACM transitions from selling > institutional "access" to an institutional "OA publication" model and more > and more of the articles published in the ACM DL are published in front of > the subscription paywall. > > A transition to fair Open Access practices would require the difficult > decision of giving up on licensing revenue. > The ACM does not seem willing to do it, and cannot be trusted to do it > eventually. > > > On Fri, Dec 20, 2019 at 7:08 PM Roberto Di Cosmo > > wrote: > > Thanks Gabriel for bringing this to this list: it was indeed shocking to > see ACM > (and many other learned societies) in the list of signatories of this > letter. > > The fact that many small learned societies do not feel ready to jump into > a pure > open access model right away does not justify their signature on a letter > containing highly debatable (that's an euphemism) statements like the ones > you pinpoint. > > By a curious coincidence, I got almost at the same time an ACM newlsetter > (Blue > Diamond) containing among other announcements, this one: > > ACM OPEN: A New Transformative Model for Open Access Publication > > Over the past year ACM Publications staff have been working > collaboratively with > a group of large research universities in the United States to > develop an > entirely new and innovative model for Open Access publication that > has the > potential to transition ACM into a predominantly Open Access > publisher over the > next decade or sooner. > > You can find details of the proposed model at > https://www.acm.org/publications/openaccess#acmopen > > Cheers > > -- > Roberto > > On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Dear types-list and SIGPLAN, > > I have long been of the opinion that our scientific publications > should be Open Access, and that editors should not request more than > a fair price (cost of publication, which Dasgtuhl estimates at $60 > per article). In particular, I believe that copyright transfer > agreements, as imposed by most editors including the ACM, is deeply > unethical: the publishers are not the authors of our scientific > production and they should not force us to give our copyright to > them. A non-exclusive publishing agreement should be enough. > > Whether or not you agree with this position, you may be interested in > the content of the following letter to the US White House that > a coalition of scientific publishers, *including the ACM*, signed and > support. > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > press release from the coalition of editors: > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > (This letter was written in the context of a proposed US legislation > to force more scientists to publish their work in Fair Open Access > venues. I haven't been able to find a precise link to a discussion of > this proposed legislation.) > > The following parts of the letter co-signed by the ACM are > particularly juicy: > > [We] have learned that the Administration may be preparing to step > into the private marketplace and force the immediate free distribution > of journal articles financed and published by organizations in the > private sector, including many non-profits. This would effectively > nationalize the valuable American intellectual property that we > produce and force us to give it away to the rest of the world for > free. > This mandate [...] would make it very difficult for most American > publishers to invest in publishing these articles. As a consequence, > it would place increased financial responsibility on the government > through diverted federal research grant funds or additional monies > to underwrite the important value added by publishing. In the coming > years, this cost shift would place billions of dollars of new and > additional burden on taxpayers. > > In my discussion with many of us, I regularly hear that the ACM is > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > its weird views (for example, that it really does cost $700 or $900 to > publish an article as Open Source) is good for our research > community. It do not see how this argument is compatible with the ACM > signing this letter. > > I believe that many of our activities, which we collectively trained > ourselves to see as harmless administrative details of our research > work, are in fact empowering the ACM to make those claims. Should we > accept to give away our copyright, or pay? unreasonable > Gold Access author processing charges (APCs)? > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > shielding many of us from paying APCs. But many of the smaller > conferences, symposiums or workshops in our community whose > proceedings are handled by the ACM are still limited to "pay > $900" (or "pay $25 per page") as the only Open Access option, with > copyright transfer as the only free choice, which is effectively > keeping those proceedings Closed-Access. > > -- > Roberto Di Cosmo > > ------------------------------------------------------------------ > Computer Science Professor > (on leave at INRIA from IRIF/University Paris Diderot) > > Director > Software Heritage https://www.softwareheritage.org > INRIA > Bureau C328 E-mail : roberto at dicosmo.org > 2, Rue Simone Iff Web page : http://www.dicosmo.org > CS 42112 Twitter : http://twitter.com/rdicosmo > 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 > ------------------------------------------------------------------ > > GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 > > > > From monnier at iro.umontreal.ca Sun Dec 22 13:15:15 2019 From: monnier at iro.umontreal.ca (Stefan Monnier) Date: Sun, 22 Dec 2019 13:15:15 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: (Andrew Myers's message of "Sun, 22 Dec 2019 09:46:15 -0500") References: <20191220180807.GA9697@traveler> Message-ID: > It feels a bit facile to bash the ACM for signing onto this letter. > The letter does not mean that they oppose making publications freely > available; in fact, I believe open access is a goal for ACM. I understand it and I would not fault ACM for opposing that new legislation (although I think it's short sighted: those rules would impose restriction to the publication-market which would likely play in favor of actors like ACM compared to more commercially minded players). But I find the letter's tone and content appalling: they should have sent *another* letter if they wanted to oppose the legislation. The ends don't justify the means. This event seems to be a good opportunity to increase pressure on the ACM to change its model, but I think we should do so in 2 separate steps (not necessarily sequential), one being about retracting the signature from that unacceptable letter, and the other about changing the aim from "expensive Gold Open Access" to a cheaper form of it (which will likely require a redesign, e.g. to distinguish "access to the DL" from "access to individual articles": a $100 author publishing fee could cover the cost of maintaining a plain archive of PDFs, while access to the richer DL and related services could be limited to ACM members (and institutions paying for the service, typically via their library)). Stefan From vitekj at icloud.com Sun Dec 22 13:18:15 2019 From: vitekj at icloud.com (jan Vitek) Date: Sun, 22 Dec 2019 13:18:15 -0500 Subject: [TYPES] Types-list Digest, Vol 121, Issue 4 In-Reply-To: References: Message-ID: <05D0D60D-9C49-427D-B6D3-41A30D42377E@icloud.com> Andrew, One problem here is that ACM is financing a whole lot of other things besides publishing our papers with the DL?s revenue. They like to call those things their ?Good Works? but I have never seen convincing accounting of what they are or an open discussion with the community if there is strong support for spending our DL revenues on them. So, when you talk about the economics ? you have to qualify this by saying ?assuming we agree with ACM on their use of our funds?. Many of us don?t. Springer gives ETAPS APCs of 200$ (if I recall). The ACM document talks about APC charges of up to 3000$. They are not reasonable. -jan Jan Vitek, Professor Computer Science, Northeastern University Video chat: https://bluejeans.com/8650208205 Book a meeting: https://calendly.com/vitekj > On Dec 22, 2019, at 12:52 PM, types-list-request at lists.seas.upenn.edu wrote: > > Send Types-list mailing list submissions to > types-list at LISTS.SEAS.UPENN.EDU > > To subscribe or unsubscribe via the World Wide Web, visit > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > or, via email, send a message with subject or body 'help' to > types-list-request at LISTS.SEAS.UPENN.EDU > > You can reach the person managing the list at > types-list-owner at LISTS.SEAS.UPENN.EDU > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Types-list digest..." > > > Today's Topics: > > 1. Re: In a letter to the US White House, ACM opposes free > distribution of peer-reviewed journal articles (Andrew Myers) > 2. Re: In a letter to the US White House, ACM opposes free > distribution of peer-reviewed journal articles (Gabriel Scherer) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Sun, 22 Dec 2019 09:46:15 -0500 > From: Andrew Myers > To: Gabriel Scherer > Cc: Types list > Subject: Re: [TYPES] In a letter to the US White House, ACM opposes > free distribution of peer-reviewed journal articles > Message-ID: > Content-Type: text/plain; charset=utf-8; format=flowed > > It feels a bit facile to bash the ACM for signing onto this letter. The > letter does not mean that they oppose making publications freely > available; in fact, I believe open access is a goal for ACM. The letter > means that they oppose having the government *mandate* that all > scientific publishers operate in this way. Exactly what the right > funding model is for scientific publications is still up in the air. > Should the government spend taxes enforcing rules whose implications we > do?not?fully?understand??I?think?not. > > The discussions I have seen about this topic seem to focus on the costs > to readers and authors while completely ignoring the economics of > publishing. I hope we can agree that publishers do provide some value in > supporting the scientific process, for example by maintaining archives > of publications for decades and across formats. That value can only be > delivered if ACM et al. have money. Where are they supposed to get it? > The old model of libraries paying ACM subscriptions is dying and is > incompatible with open access. Corporate charity is unreliable and > insufficient. The only other player with an incentive to provide money > is the authors. My understanding is that the economics are forcing ACM > to?go?in?that?direction. > > I believe ACM Is trying to be a good actor here, unlike publishers that > double-dip by extracting money from both the authors (publication fees) > and the readers (subscription fees); those publishers are doing very > well financially and generating well-earned resentment. My understanding > is that ACM does not want to double-dip. Instead, the idea is that > authors at institutions with ACM subscriptions will pay lower or no fees > for publications. That should keep the total cost to institutions under > control and hopefully approximately cost-neutral. And note that the open > access fees charged to other authors are still much lower than the > author fees charged by other publishers. The journal Nature charges > authors?$2000,?for?example,?and?it?is?not?the?high?end. > > Best, > > Andrew?Myers > > Gabriel?Scherer?wrote?on?12/21/19?6:01?AM: >> [?The?Types?Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >> >> Dear?Roberto?(and?list), >> >> The?new?ACM?Open?model?is?based?on?the?core?idea?of?saving?the?licensing >> revenue?of?the?ACM?by?shifting?costs?from?their?many?customers?(including >> in?particular?companies)?to?only?the?institutions?who?submit?the?articles. >> >> They?hope?that?the?academic?actors?that?produce?the?scientific?value?will >> also?pay?for?current?ACM?expenses.?This?model?is?completely?incompatible >> with?having?fair?Open?Access?prices?for?ACM?publications;?on?the?contrary, >> >> it?would?result?in?a?strong?total-cost?increase?for?academic?entities?that >> >> publish?in?ACM?proceedings. >> >> This?is?frankly?explained?on?the?(current?version?of)?the?ACM?Open >> documentation?page: >> https://www.acm.org/publications/openaccess#acmopen >> >>> Today,?ACM?Publications?and?the?ACM?Digital?Library?platform?are?funded?by >>> >>> selling?"read"?or?"access"?licenses?to?approximately?2,700?universities, >>> government?research?labs,?and?corporations?from?around?the?world.?The >>> income?generated?from?the?sale?of?these?licenses?[...]?is?approximately >>> $20M+?annually >>> >> The?vast?majority?of?[ACM]?articles?are?authored?by?individuals?affiliated >> >>> with?~1,000?institutions,?which?is?roughly?1/3?of?the?institutions?that >>> license??access??to?the?ACM?Digital?Library.?So,?the?main?challenge?for?ACM >>> >>> is?how?to?generate?roughly?the?same?income?from?1/3?the?number?of >>> institutions?over?the?long?term,?as?ACM?transitions?from?selling >>> institutional?"access"?to?an?institutional?"OA?publication"?model?and?more >>> >>> and?more?of?the?articles?published?in?the?ACM?DL?are?published?in?front?of >>> >>> the?subscription?paywall. >>> >> A?transition?to?fair?Open?Access?practices?would?require?the?difficult >> decision?of?giving?up?on?licensing?revenue. >> The?ACM?does?not?seem?willing?to?do?it,?and?cannot?be?trusted?to?do?it >> eventually. >> >> >> On?Fri,?Dec?20,?2019?at?7:08?PM?Roberto?Di?Cosmo >> wrote: >> >>> Thanks?Gabriel?for?bringing?this?to?this?list:?it?was?indeed?shocking?to >>> see?ACM >>> (and?many?other?learned?societies)?in?the?list?of?signatories?of?this >>> letter. >>> >>> The?fact?that?many?small?learned?societies?do?not?feel?ready?to?jump?into >>> >>> a?pure >>> open?access?model?right?away?does?not?justify?their?signature?on?a?letter >>> >>> containing?highly?debatable?(that's?an?euphemism)?statements?like?the?ones >>> >>> you?pinpoint. >>> >>> By?a?curious?coincidence,?I?got?almost?at?the?same?time?an?ACM?newlsetter >>> >>> (Blue >>> Diamond)?containing?among?other?announcements,?this?one: >>> >>> ?????ACM?OPEN:?A?New?Transformative?Model?for?Open?Access?Publication >>> >>> ??????Over?the?past?year?ACM?Publications?staff?have?been?working >>> collaboratively?with >>> ??????a?group?of?large?research?universities?in?the?United?States?to >>> develop?an >>> ??????entirely?new?and?innovative?model?for?Open?Access?publication?that >>> has?the >>> ??????potential?to?transition?ACM?into?a?predominantly?Open?Access >>> publisher?over?the >>> ??????next?decade?or?sooner. >>> >>> You?can?find?details?of?the?proposed?model?at >>> https://www.acm.org/publications/openaccess#acmopen >>> >>> Cheers >>> >>> -- >>> Roberto >>> >>> On?Fri,?Dec?20,?2019?at?02:53:05PM?+0100,?Gabriel?Scherer?wrote: >>>> [?The?Types?Forum, >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>>> Dear?types-list?and?SIGPLAN, >>>> >>>> I?have?long?been?of?the?opinion?that?our?scientific?publications >>>> should?be?Open?Access,?and?that?editors?should?not?request?more?than >>>> a?fair?price?(cost?of?publication,?which?Dasgtuhl?estimates?at?$60 >>>> per?article).?In?particular,?I?believe?that?copyright?transfer >>>> agreements,?as?imposed?by?most?editors?including?the?ACM,?is?deeply >>>> unethical:?the?publishers?are?not?the?authors?of?our?scientific >>>> production?and?they?should?not?force?us?to?give?our?copyright?to >>>> them.?A?non-exclusive?publishing?agreement?should?be?enough. >>>> >>>> Whether?or?not?you?agree?with?this?position,?you?may?be?interested?in >>>> the?content?of?the?following?letter?to?the?US?White?House?that >>>> a?coalition?of?scientific?publishers,?*including?the?ACM*,?signed?and >>>> support. >>>> >>>> >>>> >>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >>> >>>> ???press?release?from?the?coalition?of?editors: >>>> >>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >>> >>>> (This?letter?was?written?in?the?context?of?a?proposed?US?legislation >>>> to?force?more?scientists?to?publish?their?work?in?Fair?Open?Access >>>> venues.?I?haven't?been?able?to?find?a?precise?link?to?a?discussion?of >>>> this?proposed?legislation.) >>>> >>>> The?following?parts?of?the?letter?co-signed?by?the?ACM?are >>>> particularly?juicy: >>>> >>>>> [We]?have?learned?that?the?Administration?may?be?preparing?to?step >>>>> into?the?private?marketplace?and?force?the?immediate?free?distribution >>>>> of?journal?articles?financed?and?published?by?organizations?in?the >>>>> private?sector,?including?many?non-profits.?This?would?effectively >>>>> nationalize?the?valuable?American?intellectual?property?that?we >>>>> produce?and?force?us?to?give?it?away?to?the?rest?of?the?world?for >>>>> free. >>>>> This?mandate?[...]?would?make?it?very?difficult?for?most?American >>>>> publishers?to?invest?in?publishing?these?articles.?As?a?consequence, >>>>> it?would?place?increased?financial?responsibility?on?the?government >>>>> through?diverted?federal?research?grant?funds?or?additional?monies >>>>> to?underwrite?the?important?value?added?by?publishing.?In?the?coming >>>>> years,?this?cost?shift?would?place?billions?of?dollars?of?new?and >>>>> additional?burden?on?taxpayers. >>>> In?my?discussion?with?many?of?us,?I?regularly?hear?that?the?ACM?is >>>> "not?evil"?(the?SIGPLAN,?of?course,?is?pure?good!)?and?that?placating >>>> its?weird?views?(for?example,?that?it?really?does?cost?$700?or?$900?to >>>> publish?an?article?as?Open?Source)?is?good?for?our?research >>>> community.?It?do?not?see?how?this?argument?is?compatible?with?the?ACM >>>> signing?this?letter. >>>> >>>> I?believe?that?many?of?our?activities,?which?we?collectively?trained >>>> ourselves?to?see?as?harmless?administrative?details?of?our?research >>>> work,?are?in?fact?empowering?the?ACM?to?make?those?claims.?Should?we >>>> accept?to?give?away?our?copyright,?or?pay??unreasonable >>>> Gold?Access?author?processing?charges?(APCs)? >>>> >>>> ?:?The?SIGPLAN?decision?to?cover?APC?costs?for?PACMPL?articles?is >>>> shielding?many?of?us?from?paying?APCs.?But?many?of?the?smaller >>>> conferences,?symposiums?or?workshops?in?our?community?whose >>>> proceedings?are?handled?by?the?ACM?are?still?limited?to?"pay >>>> $900"?(or?"pay?$25?per?page")?as?the?only?Open?Access?option,?with >>>> copyright?transfer?as?the?only?free?choice,?which?is?effectively >>>> keeping?those?proceedings?Closed-Access. >>> -- >>> Roberto?Di?Cosmo >>> >>> ------------------------------------------------------------------ >>> Computer?Science?Professor >>> ?????????????(on?leave?at?INRIA?from?IRIF/University?Paris?Diderot) >>> >>> Director >>> Software?Heritage https://www.softwareheritage.org >>> INRIA >>> Bureau?C328??????????????????E-mail?: roberto at dicosmo.org >>> 2,?Rue?Simone?Iff??????????Web?page?: http://www.dicosmo.org >>> CS?42112????????????????????Twitter?: http://twitter.com/rdicosmo >>> 75589?Paris?Cedex?12????????????Tel?:?+33?1?80?49?44?42 >>> ------------------------------------------------------------------ >>> >>> GPG?fingerprint?2931?20CE?3A5A?5390?98EC?8BFC?FCCA?C3BE?39CB?12D3 >>> >>> > > > > ------------------------------ > > Message: 2 > Date: Sun, 22 Dec 2019 18:52:12 +0100 > From: Gabriel Scherer > To: Andrew Myers > Cc: Types list > Subject: Re: [TYPES] In a letter to the US White House, ACM opposes > free distribution of peer-reviewed journal articles > Message-ID: > > Content-Type: text/plain; charset="UTF-8" > > Dear Andrew (and list), > > I believe open access is a goal for ACM > > > This is what the ACM says, but this is not their actions suggest. Some > examples: > > 1. They signed this letter. (They defend their choice in > https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) > > 2. Events affiliated with an ACM conference, such as a workshop, are not > allowed to publish their proceedings as (fair) open-access if they wish to, > for example by publishing in ETCS or LiPICS. (I know from my experience > with the ML and OCaml workshops that ACM people check this and enforce this > rule.) > > 3. According to private communication with ETAPS organizers, the Gold Open > Access deal offered by Springer costs *less* per paper for ETAPS than the > Open Access model that SIGPLAN generously funds for PACMPL. If you're doing > worse than Springer at Open Access, you are probably not trying very hard. > > I hope we can agree that publishers do provide some value in supporting the >> scientific process, for example by maintaining archives of publications for >> decades and across formats. >> > > According to LiPICS (the fair Open Access publishing arm of Schloss > Dagstuhl), their edition/typesetting work costs 60? per article ( > https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). (In > any case, ACM outsources their edition work on proceedings to external > companies, that if I understand correctly are budgeted as part of the > conference organization, so not paid by ACM itself.) > > According to arXiv, their long-term archival platform costs <$7 per article > ( https://arxiv.org/help/support/whitepaper#21-budget ). > > On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers wrote: > >> It feels a bit facile to bash the ACM for signing onto this letter. The >> letter does not mean that they oppose making publications freely available; >> in fact, I believe open access is a goal for ACM. The letter means that >> they oppose having the government *mandate* that all scientific publishers >> operate in this way. Exactly what the right funding model is for scientific >> publications is still up in the air. Should the government spend taxes >> enforcing rules whose implications we do not fully understand? I think not. >> >> The discussions I have seen about this topic seem to focus on the costs to >> readers and authors while completely ignoring the economics of publishing. >> I hope we can agree that publishers do provide some value in supporting the >> scientific process, for example by maintaining archives of publications for >> decades and across formats. That value can only be delivered if ACM et al. >> have money. Where are they supposed to get it? The old model of libraries >> paying ACM subscriptions is dying and is incompatible with open access. >> Corporate charity is unreliable and insufficient. The only other player >> with an incentive to provide money is the authors. My understanding is that >> the economics are forcing ACM to go in that direction. >> >> I believe ACM Is trying to be a good actor here, unlike publishers that >> double-dip by extracting money from both the authors (publication fees) and >> the readers (subscription fees); those publishers are doing very well >> financially and generating well-earned resentment. My understanding is that >> ACM does not want to double-dip. Instead, the idea is that authors at >> institutions with ACM subscriptions will pay lower or no fees for >> publications. That should keep the total cost to institutions under control >> and hopefully approximately cost-neutral. And note that the open access >> fees charged to other authors are still much lower than the author fees >> charged by other publishers. The journal Nature charges >> authors $2000, for example, and it is not the high end. >> >> Best, >> >> Andrew Myers >> >> Gabriel Scherer wrote on 12/21/19 6:01 AM: >> >> [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> >> >> Dear Roberto (and list), >> >> The new ACM Open model is based on the core idea of saving the licensing >> revenue of the ACM by shifting costs from their many customers (including >> in particular companies) to only the institutions who submit the articles. >> They hope that the academic actors that produce the scientific value will >> also pay for current ACM expenses. This model is completely incompatible >> with having fair Open Access prices for ACM publications; on the contrary, >> it would result in a strong total-cost increase for academic entities that >> publish in ACM proceedings. >> >> This is frankly explained on the (current version of) the ACM Open >> documentation page: >> https://www.acm.org/publications/openaccess#acmopen >> >> Today, ACM Publications and the ACM Digital Library platform are funded by >> selling "read" or "access" licenses to approximately 2,700 universities, >> government research labs, and corporations from around the world. The >> income generated from the sale of these licenses [...] is approximately >> $20M+ annually >> >> The vast majority of [ACM] articles are authored by individuals affiliated >> >> with ~1,000 institutions, which is roughly 1/3 of the institutions that >> license ?access? to the ACM Digital Library. So, the main challenge for ACM >> >> is how to generate roughly the same income from 1/3 the number of >> institutions over the long term, as ACM transitions from selling >> institutional "access" to an institutional "OA publication" model and more >> and more of the articles published in the ACM DL are published in front of >> the subscription paywall. >> >> A transition to fair Open Access practices would require the difficult >> decision of giving up on licensing revenue. >> The ACM does not seem willing to do it, and cannot be trusted to do it >> eventually. >> >> >> On Fri, Dec 20, 2019 at 7:08 PM Roberto Di Cosmo >> >> wrote: >> >> Thanks Gabriel for bringing this to this list: it was indeed shocking to >> see ACM >> (and many other learned societies) in the list of signatories of this >> letter. >> >> The fact that many small learned societies do not feel ready to jump into >> a pure >> open access model right away does not justify their signature on a letter >> containing highly debatable (that's an euphemism) statements like the ones >> you pinpoint. >> >> By a curious coincidence, I got almost at the same time an ACM newlsetter >> (Blue >> Diamond) containing among other announcements, this one: >> >> ACM OPEN: A New Transformative Model for Open Access Publication >> >> Over the past year ACM Publications staff have been working >> collaboratively with >> a group of large research universities in the United States to >> develop an >> entirely new and innovative model for Open Access publication that >> has the >> potential to transition ACM into a predominantly Open Access >> publisher over the >> next decade or sooner. >> >> You can find details of the proposed model at >> https://www.acm.org/publications/openaccess#acmopen >> >> Cheers >> >> -- >> Roberto >> >> On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: >> >> [ The Types Forum, >> >> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >> >> Dear types-list and SIGPLAN, >> >> I have long been of the opinion that our scientific publications >> should be Open Access, and that editors should not request more than >> a fair price (cost of publication, which Dasgtuhl estimates at $60 >> per article). In particular, I believe that copyright transfer >> agreements, as imposed by most editors including the ACM, is deeply >> unethical: the publishers are not the authors of our scientific >> production and they should not force us to give our copyright to >> them. A non-exclusive publishing agreement should be enough. >> >> Whether or not you agree with this position, you may be interested in >> the content of the following letter to the US White House that >> a coalition of scientific publishers, *including the ACM*, signed and >> support. >> >> >> >> >> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >> >> press release from the coalition of editors: >> >> >> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >> >> (This letter was written in the context of a proposed US legislation >> to force more scientists to publish their work in Fair Open Access >> venues. I haven't been able to find a precise link to a discussion of >> this proposed legislation.) >> >> The following parts of the letter co-signed by the ACM are >> particularly juicy: >> >> [We] have learned that the Administration may be preparing to step >> into the private marketplace and force the immediate free distribution >> of journal articles financed and published by organizations in the >> private sector, including many non-profits. This would effectively >> nationalize the valuable American intellectual property that we >> produce and force us to give it away to the rest of the world for >> free. >> This mandate [...] would make it very difficult for most American >> publishers to invest in publishing these articles. As a consequence, >> it would place increased financial responsibility on the government >> through diverted federal research grant funds or additional monies >> to underwrite the important value added by publishing. In the coming >> years, this cost shift would place billions of dollars of new and >> additional burden on taxpayers. >> >> In my discussion with many of us, I regularly hear that the ACM is >> "not evil" (the SIGPLAN, of course, is pure good!) and that placating >> its weird views (for example, that it really does cost $700 or $900 to >> publish an article as Open Source) is good for our research >> community. It do not see how this argument is compatible with the ACM >> signing this letter. >> >> I believe that many of our activities, which we collectively trained >> ourselves to see as harmless administrative details of our research >> work, are in fact empowering the ACM to make those claims. Should we >> accept to give away our copyright, or pay? unreasonable >> Gold Access author processing charges (APCs)? >> >> ?: The SIGPLAN decision to cover APC costs for PACMPL articles is >> shielding many of us from paying APCs. But many of the smaller >> conferences, symposiums or workshops in our community whose >> proceedings are handled by the ACM are still limited to "pay >> $900" (or "pay $25 per page") as the only Open Access option, with >> copyright transfer as the only free choice, which is effectively >> keeping those proceedings Closed-Access. >> >> -- >> Roberto Di Cosmo >> >> ------------------------------------------------------------------ >> Computer Science Professor >> (on leave at INRIA from IRIF/University Paris Diderot) >> >> Director >> Software Heritage https://www.softwareheritage.org >> INRIA >> Bureau C328 E-mail : roberto at dicosmo.org >> 2, Rue Simone Iff Web page : http://www.dicosmo.org >> CS 42112 Twitter : http://twitter.com/rdicosmo >> 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 >> ------------------------------------------------------------------ >> >> GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 >> >> >> >> > > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > Types-list mailing list > Types-list at LISTS.SEAS.UPENN.EDU > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > > > ------------------------------ > > End of Types-list Digest, Vol 121, Issue 4 > ****************************************** From andru at cs.cornell.edu Sun Dec 22 14:45:39 2019 From: andru at cs.cornell.edu (Andrew Myers) Date: Sun, 22 Dec 2019 14:45:39 -0500 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: <20191220180807.GA9697@traveler> Message-ID: <641a6897-bae7-ebb6-d363-de6efaf46cea@cs.cornell.edu> The reasoning behind point #1 is "If someone doesn't think the government should pass a law prohibiting X, then they support X." To be blunt, this is a fallacy. It would be interesting to see the numbers on point #3. One should keep in mind that various publishers, including ACM, have been cutting one-time deals to reduce open access article processing fees. as they explore the Open Access options. Consequently, the charges for any one conference/journal/SIG may not be representative. Since Arxiv is currently largely supported by Cornell University along with the Simons Foundation, I appreciate the callout. But its costs are also increasing dramatically. Further, ACM does many positive things beyond archiving articles. -- Andrew Gabriel Scherer wrote on 12/22/19 12:52 PM: > Dear Andrew (and list), > > I believe open access is a goal for ACM > > > This is what the ACM says, but this is not their actions suggest. Some > examples: > > 1. They signed this letter. (They defend their choice in > https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) > > 2. Events affiliated with an ACM conference, such as a workshop, are > not allowed to publish their proceedings as (fair) open-access if they > wish to, for example by publishing in ETCS or LiPICS. (I know from my > experience with the ML and OCaml workshops that ACM people check this > and enforce this rule.) > > 3. According to private communication with ETAPS organizers, the Gold > Open Access deal offered by Springer costs *less* per paper for ETAPS > than the Open Access model that SIGPLAN generously funds for PACMPL. > If you're doing worse than Springer at Open Access, you are probably > not trying very hard. > > I hope we can agree that publishers do provide some value in > supporting the scientific process, for example by maintaining > archives of publications for decades and across formats. > > > According to LiPICS (the fair Open Access publishing arm of Schloss > Dagstuhl), their edition/typesetting work costs 60? per article ( > https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). > (In any case, ACM outsources their edition work on proceedings to > external companies, that if I understand correctly are budgeted as > part of the conference organization, so not paid by ACM itself.) > > According to arXiv, their long-term archival platform costs <$7 per > article ( https://arxiv.org/help/support/whitepaper#21-budget ). > > On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers > wrote: > > It feels a bit facile to bash the ACM for signing onto this > letter. The letter does not mean that they oppose making > publications freely available; in fact, I believe open access is a > goal for ACM. The letter means that they oppose having the > government *mandate* that all scientific publishers operate in > this way. Exactly what the right funding model is for scientific > publications is still up in the air. Should the government spend > taxes enforcing rules whose implications we > do?not?fully?understand??I?think?not. > > The discussions I have seen about this topic seem to focus on the > costs to readers and authors while completely ignoring the > economics of publishing. I hope we can agree that publishers do > provide some value in supporting the scientific process, for > example by maintaining archives of publications for decades and > across formats. That value can only be delivered if ACM et al. > have money. Where are they supposed to get it? The old model of > libraries paying ACM subscriptions is dying and is incompatible > with open access. Corporate charity is unreliable and > insufficient. The only other player with an incentive to provide > money is the authors. My understanding is that the economics are > forcing ACM to?go?in?that?direction. > > I believe ACM Is trying to be a good actor here, unlike publishers > that double-dip by extracting money from both the authors > (publication fees) and the readers (subscription fees); those > publishers are doing very well financially and generating > well-earned resentment. My understanding is that ACM does not want > to double-dip. Instead, the idea is that authors at institutions > with ACM subscriptions will pay lower or no fees for publications. > That should keep the total cost to institutions under control and > hopefully approximately cost-neutral. And note that the open > access fees charged to other authors are still much lower than the > author fees charged by other publishers. The journal Nature > charges authors?$2000,?for?example,?and?it?is?not?the?high?end. > > Best, > > Andrew?Myers > > Gabriel?Scherer?wrote?on?12/21/19?6:01?AM: >> [?The?Types?Forum, >> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >> >> Dear?Roberto?(and?list), >> >> The?new?ACM?Open?model?is?based?on?the?core?idea?of?saving?the?licensing >> >> revenue?of?the?ACM?by?shifting?costs?from?their?many?customers?(including >> >> in?particular?companies)?to?only?the?institutions?who?submit?the?articles. >> >> They?hope?that?the?academic?actors?that?produce?the?scientific?value?will >> >> also?pay?for?current?ACM?expenses.?This?model?is?completely?incompatible >> >> with?having?fair?Open?Access?prices?for?ACM?publications;?on?the?contrary, >> >> it?would?result?in?a?strong?total-cost?increase?for?academic?entities?that >> >> publish?in?ACM?proceedings. >> >> This?is?frankly?explained?on?the?(current?version?of)?the?ACM?Open >> documentation?page: >> https://www.acm.org/publications/openaccess#acmopen >> >>> Today,?ACM?Publications?and?the?ACM?Digital?Library?platform?are?funded?by >>> >>> selling?"read"?or?"access"?licenses?to?approximately?2,700?universities, >>> >>> government?research?labs,?and?corporations?from?around?the?world.?The >>> >>> income?generated?from?the?sale?of?these?licenses?[...]?is?approximately >>> >>> $20M+?annually >>> >> The?vast?majority?of?[ACM]?articles?are?authored?by?individuals?affiliated >> >>> with?~1,000?institutions,?which?is?roughly?1/3?of?the?institutions?that >>> >>> license??access??to?the?ACM?Digital?Library.?So,?the?main?challenge?for?ACM >>> >>> is?how?to?generate?roughly?the?same?income?from?1/3?the?number?of >>> institutions?over?the?long?term,?as?ACM?transitions?from?selling >>> institutional?"access"?to?an?institutional?"OA?publication"?model?and?more >>> >>> and?more?of?the?articles?published?in?the?ACM?DL?are?published?in?front?of >>> >>> the?subscription?paywall. >>> >> A?transition?to?fair?Open?Access?practices?would?require?the?difficult >> >> decision?of?giving?up?on?licensing?revenue. >> The?ACM?does?not?seem?willing?to?do?it,?and?cannot?be?trusted?to?do?it >> >> eventually. >> >> >> On?Fri,?Dec?20,?2019?at?7:08?PM?Roberto?Di?Cosmo >> >> wrote: >> >>> Thanks?Gabriel?for?bringing?this?to?this?list:?it?was?indeed?shocking?to >>> >>> see?ACM >>> (and?many?other?learned?societies)?in?the?list?of?signatories?of?this >>> >>> letter. >>> >>> The?fact?that?many?small?learned?societies?do?not?feel?ready?to?jump?into >>> >>> a?pure >>> open?access?model?right?away?does?not?justify?their?signature?on?a?letter >>> >>> containing?highly?debatable?(that's?an?euphemism)?statements?like?the?ones >>> >>> you?pinpoint. >>> >>> By?a?curious?coincidence,?I?got?almost?at?the?same?time?an?ACM?newlsetter >>> >>> (Blue >>> Diamond)?containing?among?other?announcements,?this?one: >>> >>> ?????ACM?OPEN:?A?New?Transformative?Model?for?Open?Access?Publication >>> >>> >>> ??????Over?the?past?year?ACM?Publications?staff?have?been?working >>> collaboratively?with >>> ??????a?group?of?large?research?universities?in?the?United?States?to >>> >>> develop?an >>> ??????entirely?new?and?innovative?model?for?Open?Access?publication?that >>> >>> has?the >>> ??????potential?to?transition?ACM?into?a?predominantly?Open?Access >>> publisher?over?the >>> ??????next?decade?or?sooner. >>> >>> You?can?find?details?of?the?proposed?model?at >>> https://www.acm.org/publications/openaccess#acmopen >>> >>> Cheers >>> >>> -- >>> Roberto >>> >>> On?Fri,?Dec?20,?2019?at?02:53:05PM?+0100,?Gabriel?Scherer?wrote: >>>> [?The?Types?Forum, >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>>> Dear?types-list?and?SIGPLAN, >>>> >>>> I?have?long?been?of?the?opinion?that?our?scientific?publications >>>> should?be?Open?Access,?and?that?editors?should?not?request?more?than >>>> >>>> a?fair?price?(cost?of?publication,?which?Dasgtuhl?estimates?at?$60 >>>> per?article).?In?particular,?I?believe?that?copyright?transfer >>>> agreements,?as?imposed?by?most?editors?including?the?ACM,?is?deeply >>>> >>>> unethical:?the?publishers?are?not?the?authors?of?our?scientific >>>> production?and?they?should?not?force?us?to?give?our?copyright?to >>>> them.?A?non-exclusive?publishing?agreement?should?be?enough. >>>> >>>> Whether?or?not?you?agree?with?this?position,?you?may?be?interested?in >>>> >>>> the?content?of?the?following?letter?to?the?US?White?House?that >>>> a?coalition?of?scientific?publishers,?*including?the?ACM*,?signed?and >>>> >>>> support. >>>> >>>> >>>> >>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >>> >>>> ???press?release?from?the?coalition?of?editors: >>>> >>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >>> >>>> (This?letter?was?written?in?the?context?of?a?proposed?US?legislation >>>> >>>> to?force?more?scientists?to?publish?their?work?in?Fair?Open?Access >>>> venues.?I?haven't?been?able?to?find?a?precise?link?to?a?discussion?of >>>> >>>> this?proposed?legislation.) >>>> >>>> The?following?parts?of?the?letter?co-signed?by?the?ACM?are >>>> particularly?juicy: >>>> >>>>> [We]?have?learned?that?the?Administration?may?be?preparing?to?step >>>>> >>>>> into?the?private?marketplace?and?force?the?immediate?free?distribution >>>>> >>>>> of?journal?articles?financed?and?published?by?organizations?in?the >>>>> >>>>> private?sector,?including?many?non-profits.?This?would?effectively >>>>> >>>>> nationalize?the?valuable?American?intellectual?property?that?we >>>>> produce?and?force?us?to?give?it?away?to?the?rest?of?the?world?for >>>>> free. >>>>> This?mandate?[...]?would?make?it?very?difficult?for?most?American >>>>> publishers?to?invest?in?publishing?these?articles.?As?a?consequence, >>>>> >>>>> it?would?place?increased?financial?responsibility?on?the?government >>>>> >>>>> through?diverted?federal?research?grant?funds?or?additional?monies >>>>> >>>>> to?underwrite?the?important?value?added?by?publishing.?In?the?coming >>>>> >>>>> years,?this?cost?shift?would?place?billions?of?dollars?of?new?and >>>>> additional?burden?on?taxpayers. >>>> In?my?discussion?with?many?of?us,?I?regularly?hear?that?the?ACM?is >>>> "not?evil"?(the?SIGPLAN,?of?course,?is?pure?good!)?and?that?placating >>>> >>>> its?weird?views?(for?example,?that?it?really?does?cost?$700?or?$900?to >>>> >>>> publish?an?article?as?Open?Source)?is?good?for?our?research >>>> community.?It?do?not?see?how?this?argument?is?compatible?with?the?ACM >>>> >>>> signing?this?letter. >>>> >>>> I?believe?that?many?of?our?activities,?which?we?collectively?trained >>>> >>>> ourselves?to?see?as?harmless?administrative?details?of?our?research >>>> >>>> work,?are?in?fact?empowering?the?ACM?to?make?those?claims.?Should?we >>>> >>>> accept?to?give?away?our?copyright,?or?pay??unreasonable >>>> Gold?Access?author?processing?charges?(APCs)? >>>> >>>> ?:?The?SIGPLAN?decision?to?cover?APC?costs?for?PACMPL?articles?is >>>> shielding?many?of?us?from?paying?APCs.?But?many?of?the?smaller >>>> conferences,?symposiums?or?workshops?in?our?community?whose >>>> proceedings?are?handled?by?the?ACM?are?still?limited?to?"pay >>>> $900"?(or?"pay?$25?per?page")?as?the?only?Open?Access?option,?with >>>> copyright?transfer?as?the?only?free?choice,?which?is?effectively >>>> keeping?those?proceedings?Closed-Access. >>> -- >>> Roberto?Di?Cosmo >>> >>> ------------------------------------------------------------------ >>> Computer?Science?Professor >>> ?????????????(on?leave?at?INRIA?from?IRIF/University?Paris?Diderot) >>> >>> Director >>> Software?Heritage https://www.softwareheritage.org >>> INRIA >>> Bureau?C328??????????????????E-mail?: roberto at dicosmo.org >>> >>> 2,?Rue?Simone?Iff??????????Web?page?: http://www.dicosmo.org >>> CS?42112????????????????????Twitter?: http://twitter.com/rdicosmo >>> 75589?Paris?Cedex?12????????????Tel?:?+33?1?80?49?44?42 >>> ------------------------------------------------------------------ >>> >>> GPG?fingerprint?2931?20CE?3A5A?5390?98EC?8BFC?FCCA?C3BE?39CB?12D3 >>> >>> > From tadeusz.litak at gmail.com Sun Dec 22 17:21:59 2019 From: tadeusz.litak at gmail.com (Tadeusz Litak) Date: Sun, 22 Dec 2019 23:21:59 +0100 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: <641a6897-bae7-ebb6-d363-de6efaf46cea@cs.cornell.edu> References: <20191220180807.GA9697@traveler> <641a6897-bae7-ebb6-d363-de6efaf46cea@cs.cornell.edu> Message-ID: <3101e574-dbd2-2117-7975-c2d61771793f@gmail.com> Dear all, On 22.12.19 19:18, jan Vitek wrote: > Andrew, > > One problem here is that ACM is financing a whole lot of other > things besides publishing our papers with the DL?s revenue. > They like to call those things their ?Good Works? but I have > never seen convincing accounting of what they are or an open > discussion with the community if there is strong support for > spending our DL revenues on them. > > So, when you talk about the economics ? you have to qualify this > by saying ?assuming we agree with ACM on their use of our funds?. > > Many of us don?t. It's a curious thing indeed.? The ACM used to support the ICPC (International Collegiate Programming Contest) till 2017, for example. Many people haven't noted the split and one can still hear casual references to the "ACM-ICPC". But this is not one of their "Good Works" anymore, although one would think supporting the ICPC should be right up ACM's alley. It is indeed true that no matter how lean the operation, somebody has to pay for open access in the end. I know that the LMCS had a lot of problems trying to avoid charging the authors when they had to update their publishing model. Until Jir? Ad?mek's retirement, it was ran by his group as a service to the community. Now it still involves a lot of voluntary work, some of it by my colleagues at FAU Erlangen-Nuremberg, but when choosing new servers, hosting platforms etc. potential introduction of charges seems to have been a thorny issue. Had it been a truth universally acknowledged that an author in want of open access must be in possession of 100~200 EUR/USD, some of their choices could have been easier. In the end, the opinion that the community was not ready for that yet prevailed. Perhaps some insiders would like to say more on the subject. But of course there is a huge difference between these discussed charges (I guess arXiv would be in a much better financial shape if each author would chip in, say, 15~20 USD!) and the charges levied by the ACM, where 100 USD might easily be the cost of a *single page* in major conference proceedings. This could be in fact an excellent opportunity for a broader discussion of the TCS publishing model, with its absurd reliance on conference proceedings instead of normal, community-run journals. Almost every other discipline views this as an oddity. And let's be honest: it's a pathology, for more reasons than one. Now it's becoming increasingly problematic, as some researchers avoid air travel.? Were publishing in those proceedings not a matter of scientific survival for too many people, the ACM would not be in a position to impose such exorbitant charges. Many people much more accomplished than myself have complained about this, both publicly and privately. But a change can happen only if enough heavyweights (of which I am surely not one) band together, decide on one direction and act to free the world from this monster. Best, t. On 22.12.19 20:45, Andrew Myers wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > The reasoning behind point #1 is "If someone doesn't think the government should pass a law prohibiting X, then they > support X." To be blunt, this is a fallacy. > > It would be interesting to see the numbers on point #3. One should keep in mind that various publishers, including > ACM, have been cutting one-time deals to reduce open access article processing fees. as they explore the Open Access > options. Consequently, the charges for any one conference/journal/SIG may not be representative. > > Since Arxiv is currently largely supported by Cornell University along with the Simons Foundation, I appreciate the > callout. But its costs are also increasing dramatically. Further, ACM does many positive things beyond archiving > articles. > > -- Andrew > > Gabriel Scherer wrote on 12/22/19 12:52 PM: >> Dear Andrew (and list), >> >> ??? I believe open access is a goal for ACM >> >> >> This is what the ACM says, but this is not their actions suggest. Some examples: >> >> 1. They signed this letter. (They defend their choice in >> https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) >> >> 2. Events affiliated with an ACM conference, such as a workshop, are not allowed to publish their proceedings as >> (fair) open-access if they wish to, for example by publishing in ETCS or LiPICS. (I know from my experience with the >> ML and OCaml workshops that ACM people check this and enforce this rule.) >> >> 3. According to private communication with ETAPS organizers, the Gold Open Access deal offered by Springer costs >> *less* per paper for ETAPS than the Open Access model that SIGPLAN generously funds for PACMPL. If you're doing worse >> than Springer at Open Access, you are probably not trying very hard. >> >> ??? I hope we can agree that publishers do provide some value in >> ??? supporting the scientific process, for example by maintaining >> ??? archives of publications for decades and across formats. >> >> >> According to LiPICS (the fair Open Access publishing arm of Schloss Dagstuhl), their edition/typesetting work costs >> 60? per article ( https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). (In any case, ACM outsources >> their edition work on proceedings to external companies, that if I understand correctly are budgeted as part of the >> conference organization, so not paid by ACM itself.) >> >> According to arXiv, their long-term archival platform costs <$7 per article ( >> https://arxiv.org/help/support/whitepaper#21-budget ). >> >> On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers > wrote: >> >> ??? It feels a bit facile to bash the ACM for signing onto this >> ??? letter. The letter does not mean that they oppose making >> ??? publications freely available; in fact, I believe open access is a >> ??? goal for ACM. The letter means that they oppose having the >> ??? government *mandate* that all scientific publishers operate in >> ??? this way. Exactly what the right funding model is for scientific >> ??? publications is still up in the air. Should the government spend >> ??? taxes enforcing rules whose implications we >> ??? do?not?fully?understand??I?think?not. >> >> ??? The discussions I have seen about this topic seem to focus on the >> ??? costs to readers and authors while completely ignoring the >> ??? economics of publishing. I hope we can agree that publishers do >> ??? provide some value in supporting the scientific process, for >> ??? example by maintaining archives of publications for decades and >> ??? across formats. That value can only be delivered if ACM et al. >> ??? have money. Where are they supposed to get it? The old model of >> ??? libraries paying ACM subscriptions is dying and is incompatible >> ??? with open access. Corporate charity is unreliable and >> ??? insufficient. The only other player with an incentive to provide >> ??? money is the authors. My understanding is that the economics are >> ??? forcing ACM to?go?in?that?direction. >> >> ??? I believe ACM Is trying to be a good actor here, unlike publishers >> ??? that double-dip by extracting money from both the authors >> ??? (publication fees) and the readers (subscription fees); those >> ??? publishers are doing very well financially and generating >> ??? well-earned resentment. My understanding is that ACM does not want >> ??? to double-dip. Instead, the idea is that authors at institutions >> ??? with ACM subscriptions will pay lower or no fees for publications. >> ??? That should keep the total cost to institutions under control and >> ??? hopefully approximately cost-neutral. And note that the open >> ??? access fees charged to other authors are still much lower than the >> ??? author fees charged by other publishers. The journal Nature >> ??? charges authors?$2000,?for?example,?and?it?is?not?the?high?end. >> >> ??? Best, >> >> ??? Andrew?Myers >> >> ??? Gabriel?Scherer?wrote?on?12/21/19?6:01?AM: >>> ??? [?The?Types?Forum, >>> ??? http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>> >>> ??? Dear?Roberto?(and?list), >>> >>> The?new?ACM?Open?model?is?based?on?the?core?idea?of?saving?the?licensing >>> >>> revenue?of?the?ACM?by?shifting?costs?from?their?many?customers?(including >>> >>> in?particular?companies)?to?only?the?institutions?who?submit?the?articles. >>> >>> They?hope?that?the?academic?actors?that?produce?the?scientific?value?will >>> >>> also?pay?for?current?ACM?expenses.?This?model?is?completely?incompatible >>> >>> with?having?fair?Open?Access?prices?for?ACM?publications;?on?the?contrary, >>> >>> it?would?result?in?a?strong?total-cost?increase?for?academic?entities?that >>> >>> ??? publish?in?ACM?proceedings. >>> >>> This?is?frankly?explained?on?the?(current?version?of)?the?ACM?Open >>> ??? documentation?page: >>> ??? https://www.acm.org/publications/openaccess#acmopen >>> >>>> Today,?ACM?Publications?and?the?ACM?Digital?Library?platform?are?funded?by >>>> >>>> selling?"read"?or?"access"?licenses?to?approximately?2,700?universities, >>>> >>>> government?research?labs,?and?corporations?from?around?the?world.?The >>>> >>>> income?generated?from?the?sale?of?these?licenses?[...]?is?approximately >>>> >>>> ??? $20M+?annually >>>> >>> The?vast?majority?of?[ACM]?articles?are?authored?by?individuals?affiliated >>> >>>> with?~1,000?institutions,?which?is?roughly?1/3?of?the?institutions?that >>>> >>>> license??access??to?the?ACM?Digital?Library.?So,?the?main?challenge?for?ACM >>>> >>>> is?how?to?generate?roughly?the?same?income?from?1/3?the?number?of >>>> institutions?over?the?long?term,?as?ACM?transitions?from?selling >>>> institutional?"access"?to?an?institutional?"OA?publication"?model?and?more >>>> >>>> and?more?of?the?articles?published?in?the?ACM?DL?are?published?in?front?of >>>> >>>> ??? the?subscription?paywall. >>>> >>> A?transition?to?fair?Open?Access?practices?would?require?the?difficult >>> >>> ??? decision?of?giving?up?on?licensing?revenue. >>> The?ACM?does?not?seem?willing?to?do?it,?and?cannot?be?trusted?to?do?it >>> >>> ??? eventually. >>> >>> >>> ??? On?Fri,?Dec?20,?2019?at?7:08?PM?Roberto?Di?Cosmo >>> ??? >>> ??? wrote: >>> >>>> Thanks?Gabriel?for?bringing?this?to?this?list:?it?was?indeed?shocking?to >>>> >>>> ??? see?ACM >>>> (and?many?other?learned?societies)?in?the?list?of?signatories?of?this >>>> >>>> ??? letter. >>>> >>>> The?fact?that?many?small?learned?societies?do?not?feel?ready?to?jump?into >>>> >>>> ??? a?pure >>>> open?access?model?right?away?does?not?justify?their?signature?on?a?letter >>>> >>>> containing?highly?debatable?(that's?an?euphemism)?statements?like?the?ones >>>> >>>> ??? you?pinpoint. >>>> >>>> By?a?curious?coincidence,?I?got?almost?at?the?same?time?an?ACM?newlsetter >>>> >>>> ??? (Blue >>>> ??? Diamond)?containing?among?other?announcements,?this?one: >>>> >>>> ?????ACM?OPEN:?A?New?Transformative?Model?for?Open?Access?Publication >>>> >>>> >>>> ??????Over?the?past?year?ACM?Publications?staff?have?been?working >>>> ??? collaboratively?with >>>> ??????a?group?of?large?research?universities?in?the?United?States?to >>>> >>>> ??? develop?an >>>> ??????entirely?new?and?innovative?model?for?Open?Access?publication?that >>>> >>>> ??? has?the >>>> ??????potential?to?transition?ACM?into?a?predominantly?Open?Access >>>> ??? publisher?over?the >>>> ??? ??????next?decade?or?sooner. >>>> >>>> ??? You?can?find?details?of?the?proposed?model?at >>>> ??? https://www.acm.org/publications/openaccess#acmopen >>>> >>>> ??? Cheers >>>> >>>> ??? -- ??? Roberto >>>> >>>> On?Fri,?Dec?20,?2019?at?02:53:05PM?+0100,?Gabriel?Scherer?wrote: >>>>> ??? [?The?Types?Forum, >>>> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>>>> ??? Dear?types-list?and?SIGPLAN, >>>>> >>>>> I?have?long?been?of?the?opinion?that?our?scientific?publications >>>>> should?be?Open?Access,?and?that?editors?should?not?request?more?than >>>>> >>>>> a?fair?price?(cost?of?publication,?which?Dasgtuhl?estimates?at?$60 >>>>> per?article).?In?particular,?I?believe?that?copyright?transfer >>>>> agreements,?as?imposed?by?most?editors?including?the?ACM,?is?deeply >>>>> >>>>> unethical:?the?publishers?are?not?the?authors?of?our?scientific >>>>> production?and?they?should?not?force?us?to?give?our?copyright?to >>>>> them.?A?non-exclusive?publishing?agreement?should?be?enough. >>>>> >>>>> Whether?or?not?you?agree?with?this?position,?you?may?be?interested?in >>>>> >>>>> the?content?of?the?following?letter?to?the?US?White?House?that >>>>> a?coalition?of?scientific?publishers,?*including?the?ACM*,?signed?and >>>>> >>>>> ??? support. >>>>> >>>>> >>>>> >>>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >>>> >>>>> ???press?release?from?the?coalition?of?editors: >>>>> >>>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >>>> >>>>> (This?letter?was?written?in?the?context?of?a?proposed?US?legislation >>>>> >>>>> to?force?more?scientists?to?publish?their?work?in?Fair?Open?Access >>>>> venues.?I?haven't?been?able?to?find?a?precise?link?to?a?discussion?of >>>>> >>>>> ??? this?proposed?legislation.) >>>>> >>>>> The?following?parts?of?the?letter?co-signed?by?the?ACM?are >>>>> ??? particularly?juicy: >>>>> >>>>>> [We]?have?learned?that?the?Administration?may?be?preparing?to?step >>>>>> >>>>>> into?the?private?marketplace?and?force?the?immediate?free?distribution >>>>>> >>>>>> of?journal?articles?financed?and?published?by?organizations?in?the >>>>>> >>>>>> private?sector,?including?many?non-profits.?This?would?effectively >>>>>> >>>>>> nationalize?the?valuable?American?intellectual?property?that?we >>>>>> produce?and?force?us?to?give?it?away?to?the?rest?of?the?world?for >>>>>> ??? free. >>>>>> This?mandate?[...]?would?make?it?very?difficult?for?most?American >>>>>> publishers?to?invest?in?publishing?these?articles.?As?a?consequence, >>>>>> >>>>>> it?would?place?increased?financial?responsibility?on?the?government >>>>>> >>>>>> through?diverted?federal?research?grant?funds?or?additional?monies >>>>>> >>>>>> to?underwrite?the?important?value?added?by?publishing.?In?the?coming >>>>>> >>>>>> years,?this?cost?shift?would?place?billions?of?dollars?of?new?and >>>>>> ??? additional?burden?on?taxpayers. >>>>> In?my?discussion?with?many?of?us,?I?regularly?hear?that?the?ACM?is >>>>> "not?evil"?(the?SIGPLAN,?of?course,?is?pure?good!)?and?that?placating >>>>> >>>>> its?weird?views?(for?example,?that?it?really?does?cost?$700?or?$900?to >>>>> >>>>> publish?an?article?as?Open?Source)?is?good?for?our?research >>>>> community.?It?do?not?see?how?this?argument?is?compatible?with?the?ACM >>>>> >>>>> ??? signing?this?letter. >>>>> >>>>> I?believe?that?many?of?our?activities,?which?we?collectively?trained >>>>> >>>>> ourselves?to?see?as?harmless?administrative?details?of?our?research >>>>> >>>>> work,?are?in?fact?empowering?the?ACM?to?make?those?claims.?Should?we >>>>> >>>>> accept?to?give?away?our?copyright,?or?pay??unreasonable >>>>> ??? Gold?Access?author?processing?charges?(APCs)? >>>>> >>>>> ?:?The?SIGPLAN?decision?to?cover?APC?costs?for?PACMPL?articles?is >>>>> shielding?many?of?us?from?paying?APCs.?But?many?of?the?smaller >>>>> conferences,?symposiums?or?workshops?in?our?community?whose >>>>> proceedings?are?handled?by?the?ACM?are?still?limited?to?"pay >>>>> $900"?(or?"pay?$25?per?page")?as?the?only?Open?Access?option,?with >>>>> copyright?transfer?as?the?only?free?choice,?which?is?effectively >>>>> ??? keeping?those?proceedings?Closed-Access. >>>> ??? -- ??? Roberto?Di?Cosmo >>>> >>>> ------------------------------------------------------------------ >>>> ??? Computer?Science?Professor >>>> ?????????????(on?leave?at?INRIA?from?IRIF/University?Paris?Diderot) >>>> >>>> ??? Director >>>> ??? Software?Heritage https://www.softwareheritage.org >>>> ??? INRIA >>>> ??? Bureau?C328??????????????????E-mail?: roberto at dicosmo.org >>>> ??? >>>> ??? 2,?Rue?Simone?Iff??????????Web?page?: http://www.dicosmo.org >>>> ??? CS?42112????????????????????Twitter?: http://twitter.com/rdicosmo >>>> ??? 75589?Paris?Cedex?12????????????Tel?:?+33?1?80?49?44?42 >>>> ------------------------------------------------------------------ >>>> >>>> GPG?fingerprint?2931?20CE?3A5A?5390?98EC?8BFC?FCCA?C3BE?39CB?12D3 >>>> >>>> >> > From dreyer at mpi-sws.org Sun Dec 22 17:48:25 2019 From: dreyer at mpi-sws.org (Derek Dreyer) Date: Sun, 22 Dec 2019 23:48:25 +0100 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: <641a6897-bae7-ebb6-d363-de6efaf46cea@cs.cornell.edu> References: <20191220180807.GA9697@traveler> <641a6897-bae7-ebb6-d363-de6efaf46cea@cs.cornell.edu> Message-ID: > Further, ACM does many positive things beyond archiving articles. According to Crista Lopes on Twitter (I'm not sure if she's on this list): "I studied @TheOfficialACM?s finances a few years ago, when I was Treasurer of SIGPLAN. As far as I can tell, the ACM DL paywall is a small business whose revenue serves entirely to pay the staff who works for it ? sales and support ppl. Very little of it flows elsewhere... As far as I can tell, the ACM could operate based only on conferences? revenue, and ditch the paywall entirely, and everything good would still happen ? staff and all. But that would mean getting rid of the DL staff. Whose salaries come [from] the DL paywall." Derek > > Since Arxiv is currently largely supported by Cornell University along > with the Simons Foundation, I appreciate the callout. But its costs are > also increasing dramatically. Further, ACM does many positive things > beyond archiving articles. > > -- Andrew > > Gabriel Scherer wrote on 12/22/19 12:52 PM: > > Dear Andrew (and list), > > > > I believe open access is a goal for ACM > > > > > > This is what the ACM says, but this is not their actions suggest. Some > > examples: > > > > 1. They signed this letter. (They defend their choice in > > https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) > > > > 2. Events affiliated with an ACM conference, such as a workshop, are > > not allowed to publish their proceedings as (fair) open-access if they > > wish to, for example by publishing in ETCS or LiPICS. (I know from my > > experience with the ML and OCaml workshops that ACM people check this > > and enforce this rule.) > > > > 3. According to private communication with ETAPS organizers, the Gold > > Open Access deal offered by Springer costs *less* per paper for ETAPS > > than the Open Access model that SIGPLAN generously funds for PACMPL. > > If you're doing worse than Springer at Open Access, you are probably > > not trying very hard. > > > > I hope we can agree that publishers do provide some value in > > supporting the scientific process, for example by maintaining > > archives of publications for decades and across formats. > > > > > > According to LiPICS (the fair Open Access publishing arm of Schloss > > Dagstuhl), their edition/typesetting work costs 60? per article ( > > https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). > > (In any case, ACM outsources their edition work on proceedings to > > external companies, that if I understand correctly are budgeted as > > part of the conference organization, so not paid by ACM itself.) > > > > According to arXiv, their long-term archival platform costs <$7 per > > article ( https://arxiv.org/help/support/whitepaper#21-budget ). > > > > On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers > > wrote: > > > > It feels a bit facile to bash the ACM for signing onto this > > letter. The letter does not mean that they oppose making > > publications freely available; in fact, I believe open access is a > > goal for ACM. The letter means that they oppose having the > > government *mandate* that all scientific publishers operate in > > this way. Exactly what the right funding model is for scientific > > publications is still up in the air. Should the government spend > > taxes enforcing rules whose implications we > > do not fully understand? I think not. > > > > The discussions I have seen about this topic seem to focus on the > > costs to readers and authors while completely ignoring the > > economics of publishing. I hope we can agree that publishers do > > provide some value in supporting the scientific process, for > > example by maintaining archives of publications for decades and > > across formats. That value can only be delivered if ACM et al. > > have money. Where are they supposed to get it? The old model of > > libraries paying ACM subscriptions is dying and is incompatible > > with open access. Corporate charity is unreliable and > > insufficient. The only other player with an incentive to provide > > money is the authors. My understanding is that the economics are > > forcing ACM to go in that direction. > > > > I believe ACM Is trying to be a good actor here, unlike publishers > > that double-dip by extracting money from both the authors > > (publication fees) and the readers (subscription fees); those > > publishers are doing very well financially and generating > > well-earned resentment. My understanding is that ACM does not want > > to double-dip. Instead, the idea is that authors at institutions > > with ACM subscriptions will pay lower or no fees for publications. > > That should keep the total cost to institutions under control and > > hopefully approximately cost-neutral. And note that the open > > access fees charged to other authors are still much lower than the > > author fees charged by other publishers. The journal Nature > > charges authors $2000, for example, and it is not the high end. > > > > Best, > > > > Andrew Myers > > > > Gabriel Scherer wrote on 12/21/19 6:01 AM: > >> [ The Types Forum, > >> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > >> > >> Dear Roberto (and list), > >> > >> The new ACM Open model is based on the core idea of saving the licensing > >> > >> revenue of the ACM by shifting costs from their many customers (including > >> > >> in particular companies) to only the institutions who submit the articles. > >> > >> They hope that the academic actors that produce the scientific value will > >> > >> also pay for current ACM expenses. This model is completely incompatible > >> > >> with having fair Open Access prices for ACM publications; on the contrary, > >> > >> it would result in a strong total-cost increase for academic entities that > >> > >> publish in ACM proceedings. > >> > >> This is frankly explained on the (current version of) the ACM Open > >> documentation page: > >> https://www.acm.org/publications/openaccess#acmopen > >> > >>> Today, ACM Publications and the ACM Digital Library platform are funded by > >>> > >>> selling "read" or "access" licenses to approximately 2,700 universities, > >>> > >>> government research labs, and corporations from around the world. The > >>> > >>> income generated from the sale of these licenses [...] is approximately > >>> > >>> $20M+ annually > >>> > >> The vast majority of [ACM] articles are authored by individuals affiliated > >> > >>> with ~1,000 institutions, which is roughly 1/3 of the institutions that > >>> > >>> license ?access? to the ACM Digital Library. So, the main challenge for ACM > >>> > >>> is how to generate roughly the same income from 1/3 the number of > >>> institutions over the long term, as ACM transitions from selling > >>> institutional "access" to an institutional "OA publication" model and more > >>> > >>> and more of the articles published in the ACM DL are published in front of > >>> > >>> the subscription paywall. > >>> > >> A transition to fair Open Access practices would require the difficult > >> > >> decision of giving up on licensing revenue. > >> The ACM does not seem willing to do it, and cannot be trusted to do it > >> > >> eventually. > >> > >> > >> On Fri, Dec 20, 2019 at 7:08 PM Roberto Di Cosmo > >> > >> wrote: > >> > >>> Thanks Gabriel for bringing this to this list: it was indeed shocking to > >>> > >>> see ACM > >>> (and many other learned societies) in the list of signatories of this > >>> > >>> letter. > >>> > >>> The fact that many small learned societies do not feel ready to jump into > >>> > >>> a pure > >>> open access model right away does not justify their signature on a letter > >>> > >>> containing highly debatable (that's an euphemism) statements like the ones > >>> > >>> you pinpoint. > >>> > >>> By a curious coincidence, I got almost at the same time an ACM newlsetter > >>> > >>> (Blue > >>> Diamond) containing among other announcements, this one: > >>> > >>> ACM OPEN: A New Transformative Model for Open Access Publication > >>> > >>> > >>> Over the past year ACM Publications staff have been working > >>> collaboratively with > >>> a group of large research universities in the United States to > >>> > >>> develop an > >>> entirely new and innovative model for Open Access publication that > >>> > >>> has the > >>> potential to transition ACM into a predominantly Open Access > >>> publisher over the > >>> next decade or sooner. > >>> > >>> You can find details of the proposed model at > >>> https://www.acm.org/publications/openaccess#acmopen > >>> > >>> Cheers > >>> > >>> -- > >>> Roberto > >>> > >>> On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: > >>>> [ The Types Forum, > >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > >>>> Dear types-list and SIGPLAN, > >>>> > >>>> I have long been of the opinion that our scientific publications > >>>> should be Open Access, and that editors should not request more than > >>>> > >>>> a fair price (cost of publication, which Dasgtuhl estimates at $60 > >>>> per article). In particular, I believe that copyright transfer > >>>> agreements, as imposed by most editors including the ACM, is deeply > >>>> > >>>> unethical: the publishers are not the authors of our scientific > >>>> production and they should not force us to give our copyright to > >>>> them. A non-exclusive publishing agreement should be enough. > >>>> > >>>> Whether or not you agree with this position, you may be interested in > >>>> > >>>> the content of the following letter to the US White House that > >>>> a coalition of scientific publishers, *including the ACM*, signed and > >>>> > >>>> support. > >>>> > >>>> > >>>> > >>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > >>> > >>>> press release from the coalition of editors: > >>>> > >>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > >>> > >>>> (This letter was written in the context of a proposed US legislation > >>>> > >>>> to force more scientists to publish their work in Fair Open Access > >>>> venues. I haven't been able to find a precise link to a discussion of > >>>> > >>>> this proposed legislation.) > >>>> > >>>> The following parts of the letter co-signed by the ACM are > >>>> particularly juicy: > >>>> > >>>>> [We] have learned that the Administration may be preparing to step > >>>>> > >>>>> into the private marketplace and force the immediate free distribution > >>>>> > >>>>> of journal articles financed and published by organizations in the > >>>>> > >>>>> private sector, including many non-profits. This would effectively > >>>>> > >>>>> nationalize the valuable American intellectual property that we > >>>>> produce and force us to give it away to the rest of the world for > >>>>> free. > >>>>> This mandate [...] would make it very difficult for most American > >>>>> publishers to invest in publishing these articles. As a consequence, > >>>>> > >>>>> it would place increased financial responsibility on the government > >>>>> > >>>>> through diverted federal research grant funds or additional monies > >>>>> > >>>>> to underwrite the important value added by publishing. In the coming > >>>>> > >>>>> years, this cost shift would place billions of dollars of new and > >>>>> additional burden on taxpayers. > >>>> In my discussion with many of us, I regularly hear that the ACM is > >>>> "not evil" (the SIGPLAN, of course, is pure good!) and that placating > >>>> > >>>> its weird views (for example, that it really does cost $700 or $900 to > >>>> > >>>> publish an article as Open Source) is good for our research > >>>> community. It do not see how this argument is compatible with the ACM > >>>> > >>>> signing this letter. > >>>> > >>>> I believe that many of our activities, which we collectively trained > >>>> > >>>> ourselves to see as harmless administrative details of our research > >>>> > >>>> work, are in fact empowering the ACM to make those claims. Should we > >>>> > >>>> accept to give away our copyright, or pay? unreasonable > >>>> Gold Access author processing charges (APCs)? > >>>> > >>>> ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > >>>> shielding many of us from paying APCs. But many of the smaller > >>>> conferences, symposiums or workshops in our community whose > >>>> proceedings are handled by the ACM are still limited to "pay > >>>> $900" (or "pay $25 per page") as the only Open Access option, with > >>>> copyright transfer as the only free choice, which is effectively > >>>> keeping those proceedings Closed-Access. > >>> -- > >>> Roberto Di Cosmo > >>> > >>> ------------------------------------------------------------------ > >>> Computer Science Professor > >>> (on leave at INRIA from IRIF/University Paris Diderot) > >>> > >>> Director > >>> Software Heritage https://www.softwareheritage.org > >>> INRIA > >>> Bureau C328 E-mail : roberto at dicosmo.org > >>> > >>> 2, Rue Simone Iff Web page : http://www.dicosmo.org > >>> CS 42112 Twitter : http://twitter.com/rdicosmo > >>> 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 > >>> ------------------------------------------------------------------ > >>> > >>> GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 > >>> > >>> > > > From lopes at uci.edu Sun Dec 22 18:19:44 2019 From: lopes at uci.edu (Cristina Videira Lopes) Date: Sun, 22 Dec 2019 15:19:44 -0800 Subject: [TYPES] In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles In-Reply-To: References: <20191220180807.GA9697@traveler> <641a6897-bae7-ebb6-d363-de6efaf46cea@cs.cornell.edu> Message-ID: Hi, I'm not on that list. And yes, I studied it. My conclusion at the time was that the little money that the DL generates, that is not to pay the sales and support staff who works for the DL, is for supporting the publication of journals, which don't have revenue. This was a ballpark, reverse engineering assessment by looking at the ACM tax returns and the letter that John White sent us explaining the publication costs. Unfortunately, the DL ties up the conference papers (and now the PACM papers) with the ACM journals. The costs of the former are almost 100% covered by conference registration fees. And yet, they get hidden behind the paywall, because the ACM sells the DL-as-product to the libraries, and the more content, the more valuable it is. The ACM would have a very weak product if it sold only the journals. Best, Crista On 12/22/2019 2:48 PM, Derek Dreyer wrote: >> Further, ACM does many positive things beyond archiving articles. > > According to Crista Lopes on Twitter (I'm not sure if she's on this list): > > "I studied @TheOfficialACM?s finances a few years ago, when I was > Treasurer of SIGPLAN. As far as I can tell, the ACM DL paywall is a > small business whose revenue serves entirely to pay the staff who > works for it ? sales and support ppl. Very little of it flows > elsewhere... As far as I can tell, the ACM could operate based only > on conferences? revenue, and ditch the paywall entirely, and > everything good would still happen ? staff and all. But that would > mean getting rid of the DL staff. Whose salaries come [from] the DL > paywall." > > Derek > >> >> Since Arxiv is currently largely supported by Cornell University along >> with the Simons Foundation, I appreciate the callout. But its costs are >> also increasing dramatically. Further, ACM does many positive things >> beyond archiving articles. >> >> -- Andrew >> >> Gabriel Scherer wrote on 12/22/19 12:52 PM: >>> Dear Andrew (and list), >>> >>> I believe open access is a goal for ACM >>> >>> >>> This is what the ACM says, but this is not their actions suggest. Some >>> examples: >>> >>> 1. They signed this letter. (They defend their choice in >>> https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) >>> >>> 2. Events affiliated with an ACM conference, such as a workshop, are >>> not allowed to publish their proceedings as (fair) open-access if they >>> wish to, for example by publishing in ETCS or LiPICS. (I know from my >>> experience with the ML and OCaml workshops that ACM people check this >>> and enforce this rule.) >>> >>> 3. According to private communication with ETAPS organizers, the Gold >>> Open Access deal offered by Springer costs *less* per paper for ETAPS >>> than the Open Access model that SIGPLAN generously funds for PACMPL. >>> If you're doing worse than Springer at Open Access, you are probably >>> not trying very hard. >>> >>> I hope we can agree that publishers do provide some value in >>> supporting the scientific process, for example by maintaining >>> archives of publications for decades and across formats. >>> >>> >>> According to LiPICS (the fair Open Access publishing arm of Schloss >>> Dagstuhl), their edition/typesetting work costs 60? per article ( >>> https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). >>> (In any case, ACM outsources their edition work on proceedings to >>> external companies, that if I understand correctly are budgeted as >>> part of the conference organization, so not paid by ACM itself.) >>> >>> According to arXiv, their long-term archival platform costs <$7 per >>> article ( https://arxiv.org/help/support/whitepaper#21-budget ). >>> >>> On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers >> > wrote: >>> >>> It feels a bit facile to bash the ACM for signing onto this >>> letter. The letter does not mean that they oppose making >>> publications freely available; in fact, I believe open access is a >>> goal for ACM. The letter means that they oppose having the >>> government *mandate* that all scientific publishers operate in >>> this way. Exactly what the right funding model is for scientific >>> publications is still up in the air. Should the government spend >>> taxes enforcing rules whose implications we >>> do not fully understand? I think not. >>> >>> The discussions I have seen about this topic seem to focus on the >>> costs to readers and authors while completely ignoring the >>> economics of publishing. I hope we can agree that publishers do >>> provide some value in supporting the scientific process, for >>> example by maintaining archives of publications for decades and >>> across formats. That value can only be delivered if ACM et al. >>> have money. Where are they supposed to get it? The old model of >>> libraries paying ACM subscriptions is dying and is incompatible >>> with open access. Corporate charity is unreliable and >>> insufficient. The only other player with an incentive to provide >>> money is the authors. My understanding is that the economics are >>> forcing ACM to go in that direction. >>> >>> I believe ACM Is trying to be a good actor here, unlike publishers >>> that double-dip by extracting money from both the authors >>> (publication fees) and the readers (subscription fees); those >>> publishers are doing very well financially and generating >>> well-earned resentment. My understanding is that ACM does not want >>> to double-dip. Instead, the idea is that authors at institutions >>> with ACM subscriptions will pay lower or no fees for publications. >>> That should keep the total cost to institutions under control and >>> hopefully approximately cost-neutral. And note that the open >>> access fees charged to other authors are still much lower than the >>> author fees charged by other publishers. The journal Nature >>> charges authors $2000, for example, and it is not the high end. >>> >>> Best, >>> >>> Andrew Myers >>> >>> Gabriel Scherer wrote on 12/21/19 6:01 AM: >>>> [ The Types Forum, >>>> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >>>> >>>> Dear Roberto (and list), >>>> >>>> The new ACM Open model is based on the core idea of saving the licensing >>>> >>>> revenue of the ACM by shifting costs from their many customers (including >>>> >>>> in particular companies) to only the institutions who submit the articles. >>>> >>>> They hope that the academic actors that produce the scientific value will >>>> >>>> also pay for current ACM expenses. This model is completely incompatible >>>> >>>> with having fair Open Access prices for ACM publications; on the contrary, >>>> >>>> it would result in a strong total-cost increase for academic entities that >>>> >>>> publish in ACM proceedings. >>>> >>>> This is frankly explained on the (current version of) the ACM Open >>>> documentation page: >>>> https://www.acm.org/publications/openaccess#acmopen >>>> >>>>> Today, ACM Publications and the ACM Digital Library platform are funded by >>>>> >>>>> selling "read" or "access" licenses to approximately 2,700 universities, >>>>> >>>>> government research labs, and corporations from around the world. The >>>>> >>>>> income generated from the sale of these licenses [...] is approximately >>>>> >>>>> $20M+ annually >>>>> >>>> The vast majority of [ACM] articles are authored by individuals affiliated >>>> >>>>> with ~1,000 institutions, which is roughly 1/3 of the institutions that >>>>> >>>>> license ?access? to the ACM Digital Library. So, the main challenge for ACM >>>>> >>>>> is how to generate roughly the same income from 1/3 the number of >>>>> institutions over the long term, as ACM transitions from selling >>>>> institutional "access" to an institutional "OA publication" model and more >>>>> >>>>> and more of the articles published in the ACM DL are published in front of >>>>> >>>>> the subscription paywall. >>>>> >>>> A transition to fair Open Access practices would require the difficult >>>> >>>> decision of giving up on licensing revenue. >>>> The ACM does not seem willing to do it, and cannot be trusted to do it >>>> >>>> eventually. >>>> >>>> >>>> On Fri, Dec 20, 2019 at 7:08 PM Roberto Di Cosmo >>>> >>>> wrote: >>>> >>>>> Thanks Gabriel for bringing this to this list: it was indeed shocking to >>>>> >>>>> see ACM >>>>> (and many other learned societies) in the list of signatories of this >>>>> >>>>> letter. >>>>> >>>>> The fact that many small learned societies do not feel ready to jump into >>>>> >>>>> a pure >>>>> open access model right away does not justify their signature on a letter >>>>> >>>>> containing highly debatable (that's an euphemism) statements like the ones >>>>> >>>>> you pinpoint. >>>>> >>>>> By a curious coincidence, I got almost at the same time an ACM newlsetter >>>>> >>>>> (Blue >>>>> Diamond) containing among other announcements, this one: >>>>> >>>>> ACM OPEN: A New Transformative Model for Open Access Publication >>>>> >>>>> >>>>> Over the past year ACM Publications staff have been working >>>>> collaboratively with >>>>> a group of large research universities in the United States to >>>>> >>>>> develop an >>>>> entirely new and innovative model for Open Access publication that >>>>> >>>>> has the >>>>> potential to transition ACM into a predominantly Open Access >>>>> publisher over the >>>>> next decade or sooner. >>>>> >>>>> You can find details of the proposed model at >>>>> https://www.acm.org/publications/openaccess#acmopen >>>>> >>>>> Cheers >>>>> >>>>> -- >>>>> Roberto >>>>> >>>>> On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: >>>>>> [ The Types Forum, >>>>> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >>>>>> Dear types-list and SIGPLAN, >>>>>> >>>>>> I have long been of the opinion that our scientific publications >>>>>> should be Open Access, and that editors should not request more than >>>>>> >>>>>> a fair price (cost of publication, which Dasgtuhl estimates at $60 >>>>>> per article). In particular, I believe that copyright transfer >>>>>> agreements, as imposed by most editors including the ACM, is deeply >>>>>> >>>>>> unethical: the publishers are not the authors of our scientific >>>>>> production and they should not force us to give our copyright to >>>>>> them. A non-exclusive publishing agreement should be enough. >>>>>> >>>>>> Whether or not you agree with this position, you may be interested in >>>>>> >>>>>> the content of the following letter to the US White House that >>>>>> a coalition of scientific publishers, *including the ACM*, signed and >>>>>> >>>>>> support. >>>>>> >>>>>> >>>>>> >>>>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >>>>> >>>>>> press release from the coalition of editors: >>>>>> >>>>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >>>>> >>>>>> (This letter was written in the context of a proposed US legislation >>>>>> >>>>>> to force more scientists to publish their work in Fair Open Access >>>>>> venues. I haven't been able to find a precise link to a discussion of >>>>>> >>>>>> this proposed legislation.) >>>>>> >>>>>> The following parts of the letter co-signed by the ACM are >>>>>> particularly juicy: >>>>>> >>>>>>> [We] have learned that the Administration may be preparing to step >>>>>>> >>>>>>> into the private marketplace and force the immediate free distribution >>>>>>> >>>>>>> of journal articles financed and published by organizations in the >>>>>>> >>>>>>> private sector, including many non-profits. This would effectively >>>>>>> >>>>>>> nationalize the valuable American intellectual property that we >>>>>>> produce and force us to give it away to the rest of the world for >>>>>>> free. >>>>>>> This mandate [...] would make it very difficult for most American >>>>>>> publishers to invest in publishing these articles. As a consequence, >>>>>>> >>>>>>> it would place increased financial responsibility on the government >>>>>>> >>>>>>> through diverted federal research grant funds or additional monies >>>>>>> >>>>>>> to underwrite the important value added by publishing. In the coming >>>>>>> >>>>>>> years, this cost shift would place billions of dollars of new and >>>>>>> additional burden on taxpayers. >>>>>> In my discussion with many of us, I regularly hear that the ACM is >>>>>> "not evil" (the SIGPLAN, of course, is pure good!) and that placating >>>>>> >>>>>> its weird views (for example, that it really does cost $700 or $900 to >>>>>> >>>>>> publish an article as Open Source) is good for our research >>>>>> community. It do not see how this argument is compatible with the ACM >>>>>> >>>>>> signing this letter. >>>>>> >>>>>> I believe that many of our activities, which we collectively trained >>>>>> >>>>>> ourselves to see as harmless administrative details of our research >>>>>> >>>>>> work, are in fact empowering the ACM to make those claims. Should we >>>>>> >>>>>> accept to give away our copyright, or pay? unreasonable >>>>>> Gold Access author processing charges (APCs)? >>>>>> >>>>>> ?: The SIGPLAN decision to cover APC costs for PACMPL articles is >>>>>> shielding many of us from paying APCs. But many of the smaller >>>>>> conferences, symposiums or workshops in our community whose >>>>>> proceedings are handled by the ACM are still limited to "pay >>>>>> $900" (or "pay $25 per page") as the only Open Access option, with >>>>>> copyright transfer as the only free choice, which is effectively >>>>>> keeping those proceedings Closed-Access. >>>>> -- >>>>> Roberto Di Cosmo >>>>> >>>>> ------------------------------------------------------------------ >>>>> Computer Science Professor >>>>> (on leave at INRIA from IRIF/University Paris Diderot) >>>>> >>>>> Director >>>>> Software Heritage https://www.softwareheritage.org >>>>> INRIA >>>>> Bureau C328 E-mail : roberto at dicosmo.org >>>>> >>>>> 2, Rue Simone Iff Web page : http://www.dicosmo.org >>>>> CS 42112 Twitter : http://twitter.com/rdicosmo >>>>> 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 >>>>> ------------------------------------------------------------------ >>>>> >>>>> GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 >>>>> >>>>> >>> >> > -- Cristina Videira Lopes Professor of Software Engineering Master of Software Engineering Faculty Director Bren School of Information and Computer Sciences University of California, Irvine From rnrand at gmail.com Sun Dec 22 21:11:42 2019 From: rnrand at gmail.com (Robert Rand) Date: Sun, 22 Dec 2019 21:11:42 -0500 Subject: [TYPES] In a letter to the US White House, In-Reply-To: References: Message-ID: I agree with Peter and Gabriel. The ACM seems intent on charging the same amount per article and just shifting around who pays for it: from readers to authors to conference-goers to (in their newest scheme) universities as part of some sort of general tax. And we shouldn't be paying it. I would add that for publishing conference proceedings, EPTCS (which publishes the conference proceedings for Linearity, ICLP, QPL and dozens of other conferences) is a great option. EPTCS is free to everyone, and it would be nice if more programming languages conferences switched to using it. On Sun, Dec 22, 2019 at 12:38 PM wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list > ] > > I was recently invited to the editorial board of a new ACM journal. I > refused to become an editor, on the grounds that the journal is not > open access. > > I feel that as editors, we volunteer our time, and I have decided to > do so only for journals that are truly open access. And by "open > access", I do not mean some "gold" model based on author charges > (currently $1300-$1700 per ACM journal article, see > https://www.acm.org/publications/openaccess). I mean journals that are > free for both authors and readers. > > There are already many such journals, and they are usually > community-run. Examples include Logical Methods in Computer Science > (https://lmcs.episciences.org/), Compositionality > (https://www.compositionality-journal.org/), and Quantum > (https://quantum-journal.org/). There is relatively little in the way > of interested groups of individuals starting new such journals. > > I think the best way to get rid of unreasonable publishing models is > not to participate in them. > > -- Peter > > Jonathan Aldrich wrote: > > > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > gabriel.scherer at gmail.com> > > wrote: > > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > ] > > > > > > Dear types-list and SIGPLAN, > > > > > > I have long been of the opinion that our scientific publications > > > should be Open Access, and that editors should not request more than > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > per article). In particular, I believe that copyright transfer > > > agreements, as imposed by most editors including the ACM, is deeply > > > unethical: the publishers are not the authors of our scientific > > > production and they should not force us to give our copyright to > > > them. A non-exclusive publishing agreement should be enough. > > > > > > Whether or not you agree with this position, you may be interested in > > > the content of the following letter to the US White House that > > > a coalition of scientific publishers, *including the ACM*, signed and > > > support. > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > press release from the coalition of editors: > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > (This letter was written in the context of a proposed US legislation > > > to force more scientists to publish their work in Fair Open Access > > > venues. I haven't been able to find a precise link to a discussion of > > > this proposed legislation.) > > > > > > The following parts of the letter co-signed by the ACM are > > > particularly juicy: > > > > > > > [We] have learned that the Administration may be preparing to step > > > > into the private marketplace and force the immediate free > distribution > > > > of journal articles financed and published by organizations in the > > > > private sector, including many non-profits. This would effectively > > > > nationalize the valuable American intellectual property that we > > > > produce and force us to give it away to the rest of the world for > > > > free. > > > > > > > This mandate [...] would make it very difficult for most American > > > > publishers to invest in publishing these articles. As a consequence, > > > > it would place increased financial responsibility on the government > > > > through diverted federal research grant funds or additional monies > > > > to underwrite the important value added by publishing. In the coming > > > > years, this cost shift would place billions of dollars of new and > > > > additional burden on taxpayers. > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > its weird views (for example, that it really does cost $700 or $900 to > > > publish an article as Open Source) is good for our research > > > community. It do not see how this argument is compatible with the ACM > > > signing this letter. > > > > > > I believe that many of our activities, which we collectively trained > > > ourselves to see as harmless administrative details of our research > > > work, are in fact empowering the ACM to make those claims. Should we > > > accept to give away our copyright, or pay? unreasonable > > > Gold Access author processing charges (APCs)? > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > shielding many of us from paying APCs. But many of the smaller > > > conferences, symposiums or workshops in our community whose > > > proceedings are handled by the ACM are still limited to "pay > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > copyright transfer as the only free choice, which is effectively > > > keeping those proceedings Closed-Access. > > > > > > > > > From gabriel.scherer at gmail.com Mon Dec 23 10:28:35 2019 From: gabriel.scherer at gmail.com (Gabriel Scherer) Date: Mon, 23 Dec 2019 16:28:35 +0100 Subject: [TYPES] In a letter to the US White House, In-Reply-To: References: Message-ID: I strongly agree with Peter: the most effective action against unreasonable publishing models is to not participate in them. We are very fortunate to work in a community where many sensible Open Access venues exist, so that this decision comes at little to no personal cost. I created a webpage to document this policy (tentatively named "Keeping my work Open"). To evaluate the impact on my work since I started applying it in 2017, I listed venues to which I did or did not participate due to the policy. http://gasche.info/open_access.html The summary is that this principle is not limiting: most venues in Programming Languages research have sensible policies (I include PACMPL, thanks to the generous support of SIGPLAN to fund APCs, and ETAPS since 2018), and that the Closed Access venues I encountered had fairly direct replacements in all cases. On Sun, Dec 22, 2019 at 7:37 AM wrote: > I was recently invited to the editorial board of a new ACM journal. I > refused to become an editor, on the grounds that the journal is not > open access. > > I feel that as editors, we volunteer our time, and I have decided to > do so only for journals that are truly open access. And by "open > access", I do not mean some "gold" model based on author charges > (currently $1300-$1700 per ACM journal article, see > https://www.acm.org/publications/openaccess). I mean journals that are > free for both authors and readers. > > There are already many such journals, and they are usually > community-run. Examples include Logical Methods in Computer Science > (https://lmcs.episciences.org/), Compositionality > (https://www.compositionality-journal.org/), and Quantum > (https://quantum-journal.org/). There is relatively little in the way > of interested groups of individuals starting new such journals. > > I think the best way to get rid of unreasonable publishing models is > not to participate in them. > > -- Peter > > Jonathan Aldrich wrote: > > > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > gabriel.scherer at gmail.com> > > wrote: > > > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > ] > > > > > > Dear types-list and SIGPLAN, > > > > > > I have long been of the opinion that our scientific publications > > > should be Open Access, and that editors should not request more than > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > per article). In particular, I believe that copyright transfer > > > agreements, as imposed by most editors including the ACM, is deeply > > > unethical: the publishers are not the authors of our scientific > > > production and they should not force us to give our copyright to > > > them. A non-exclusive publishing agreement should be enough. > > > > > > Whether or not you agree with this position, you may be interested in > > > the content of the following letter to the US White House that > > > a coalition of scientific publishers, *including the ACM*, signed and > > > support. > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > press release from the coalition of editors: > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > (This letter was written in the context of a proposed US legislation > > > to force more scientists to publish their work in Fair Open Access > > > venues. I haven't been able to find a precise link to a discussion of > > > this proposed legislation.) > > > > > > The following parts of the letter co-signed by the ACM are > > > particularly juicy: > > > > > > > [We] have learned that the Administration may be preparing to step > > > > into the private marketplace and force the immediate free > distribution > > > > of journal articles financed and published by organizations in the > > > > private sector, including many non-profits. This would effectively > > > > nationalize the valuable American intellectual property that we > > > > produce and force us to give it away to the rest of the world for > > > > free. > > > > > > > This mandate [...] would make it very difficult for most American > > > > publishers to invest in publishing these articles. As a consequence, > > > > it would place increased financial responsibility on the government > > > > through diverted federal research grant funds or additional monies > > > > to underwrite the important value added by publishing. In the coming > > > > years, this cost shift would place billions of dollars of new and > > > > additional burden on taxpayers. > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > its weird views (for example, that it really does cost $700 or $900 to > > > publish an article as Open Source) is good for our research > > > community. It do not see how this argument is compatible with the ACM > > > signing this letter. > > > > > > I believe that many of our activities, which we collectively trained > > > ourselves to see as harmless administrative details of our research > > > work, are in fact empowering the ACM to make those claims. Should we > > > accept to give away our copyright, or pay? unreasonable > > > Gold Access author processing charges (APCs)? > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > shielding many of us from paying APCs. But many of the smaller > > > conferences, symposiums or workshops in our community whose > > > proceedings are handled by the ACM are still limited to "pay > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > copyright transfer as the only free choice, which is effectively > > > keeping those proceedings Closed-Access. > > > > > > > > > From roberto at dicosmo.org Tue Dec 24 12:21:38 2019 From: roberto at dicosmo.org (Roberto Di Cosmo) Date: Tue, 24 Dec 2019 18:21:38 +0100 Subject: [TYPES] Open Access: a bit of background (Was: Re: In a letter to the US White House, ACM opposes free distribution of peer-reviewed journal articles) Message-ID: <20191224172138.GC32219@traveler> Dear all, I'd like to contribute to this very interesting thread by offering some background on the Open Acces debate, that started almost a quarter of a century ago, and that may help younger collegues to get an idea of why we ended up where we are today. It's important to recall that, no matter the official statements one can read here and there, the goals of researchers on one side, and traditional scientific publishers on the other, *have always been quite different*: researchers wanted access to publications as broad and efficient as possible, and a means to signal excellence of their work; publishers wanted to run a healthy business, and a monopoly on their own areas (via mandatory copyright transfer and ownership of the publication trademarks). It just turns out that in the pre-Internet era what publishers offered also satisfied the goals of researchers: the difficult divorce started some twenty years ago, with the generalisation of the Internet, when sharing papers on the Web became a *more efficient* means of distibution than having copies of journals or proceedings sent to thousands of libraries all over the world, and the publishers' own goals started to become an obstacle to the free dissemination of science. Unfortunately, before this divorce started, northern emisphere researchers (*), as well as funding agencies, had broadly agreed to the conditions imposed by publishers, in particular mandatory transfer to them of exclusive copyright, which were not necessary to achieve the researcher's goals. This had quite disastrous consequences. On the one hand, it created a gigantic legacy problem: even if we all started publishing Open Access today, what about the hunderds of thousands of articles published over the past decades, that are still behind a paywall? On the other hand, and maybe even worse, it accustomed all the players (including governments and funding agencies) to accept the idea that the publishing houses actually *own* the copyright to the articles we write, while they actually *extort* this copyright from the authors themselves, by forcing them to relinquish their rights in order to be published, a provision that is in direct violation of the spirit of copyright itself (in France, it is actually in violation of the letter of copyright law, see article L.131-4 of the CPI). You can find a trace of this line of thinking in the wording used in the letter that started this whole thread, where it mentions "intellectual property" that belongs to the publishers. A broader and longer analysis of what is at stake can be found in this early account I wrote 15 years ago, when some of us had the illusion that Open Access was going to win soon : Scientific Publications: The Role of Public Administrations in The ICT Era, Upgrade, 2006 + available at: http://www.dicosmo.org/FSP/FATOS-Upgrade-03-2006.pdf + french original version available at http://www.dicosmo.org/FSP/FreeAccessToScience.pdf Cheers -- Roberto (*) in Latin America the situation is quite different ------------------------------------------------------------------ Computer Science Professor (on leave at INRIA from IRIF/University Paris Diderot) Director Software Heritage https://www.softwareheritage.org INRIA Bureau C328 E-mail : roberto at dicosmo.org 2, Rue Simone Iff Web page : http://www.dicosmo.org CS 42112 Twitter : http://twitter.com/rdicosmo 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 ------------------------------------------------------------------ GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 From andru at cs.cornell.edu Tue Dec 24 14:16:54 2019 From: andru at cs.cornell.edu (Andrew Myers) Date: Tue, 24 Dec 2019 14:16:54 -0500 Subject: [TYPES] Types-list Digest, Vol 121, Issue 4 In-Reply-To: <05D0D60D-9C49-427D-B6D3-41A30D42377E@icloud.com> References: <05D0D60D-9C49-427D-B6D3-41A30D42377E@icloud.com> Message-ID: <69d2366a-93f2-c07a-cc0d-e89b68e08a7f@cs.cornell.edu> Jan, I agree that it would be good for ACM to be more transparent about where the money is going, or improve their PR. I really doubt that they are being dishonest; maybe they are being inefficient or pointless activities, but I have yet to see a convincing analysis of that. I do wonder what specifically you feel they are wasting money on. I feel we tend to take the existence of professional societies like ACM and IEEE for granted, but we should be less cavalier about the possibility of blowing up their funding model. You must have misread the ACM page. The $3k APC was cited not as a "reasonable" charge, but exactly to argue that ACM is better than many commercial publishers. An APC of $3-5k is not unusual (even for some non-open-access journals!), whereas ACM's are much lower. Springer's APC vary by venue. Randomly visiting some open-access Springer journal sites suggests that the APCs for their OA journals are pretty close to those of ACM: around $1200. Andrew jan Vitek wrote on 12/22/19 1:18 PM: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > Andrew, > > One problem here is that ACM is financing a whole lot of other > things besides publishing our papers with the DL?s revenue. > They like to call those things their ?Good Works? but I have > never seen convincing accounting of what they are or an open > discussion with the community if there is strong support for > spending our DL revenues on them. > > So, when you talk about the economics ? you have to qualify this > by saying ?assuming we agree with ACM on their use of our funds?. > > Many of us don?t. > > Springer gives ETAPS APCs of 200$ (if I recall). The ACM document > talks about APC charges of up to 3000$. They are not reasonable. > > -jan > > > Jan Vitek, Professor > Computer Science, Northeastern University > Video chat: https://bluejeans.com/8650208205 > Book a meeting: https://calendly.com/vitekj > >> On Dec 22, 2019, at 12:52 PM, types-list-request at lists.seas.upenn.edu wrote: >> >> Send Types-list mailing list submissions to >> types-list at LISTS.SEAS.UPENN.EDU >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list >> or, via email, send a message with subject or body 'help' to >> types-list-request at LISTS.SEAS.UPENN.EDU >> >> You can reach the person managing the list at >> types-list-owner at LISTS.SEAS.UPENN.EDU >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Types-list digest..." >> >> >> Today's Topics: >> >> 1. Re: In a letter to the US White House, ACM opposes free >> distribution of peer-reviewed journal articles (Andrew Myers) >> 2. Re: In a letter to the US White House, ACM opposes free >> distribution of peer-reviewed journal articles (Gabriel Scherer) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Sun, 22 Dec 2019 09:46:15 -0500 >> From: Andrew Myers >> To: Gabriel Scherer >> Cc: Types list >> Subject: Re: [TYPES] In a letter to the US White House, ACM opposes >> free distribution of peer-reviewed journal articles >> Message-ID: >> Content-Type: text/plain; charset=utf-8; format=flowed >> >> It feels a bit facile to bash the ACM for signing onto this letter. The >> letter does not mean that they oppose making publications freely >> available; in fact, I believe open access is a goal for ACM. The letter >> means that they oppose having the government *mandate* that all >> scientific publishers operate in this way. Exactly what the right >> funding model is for scientific publications is still up in the air. >> Should the government spend taxes enforcing rules whose implications we >> do?not?fully?understand??I?think?not. >> >> The discussions I have seen about this topic seem to focus on the costs >> to readers and authors while completely ignoring the economics of >> publishing. I hope we can agree that publishers do provide some value in >> supporting the scientific process, for example by maintaining archives >> of publications for decades and across formats. That value can only be >> delivered if ACM et al. have money. Where are they supposed to get it? >> The old model of libraries paying ACM subscriptions is dying and is >> incompatible with open access. Corporate charity is unreliable and >> insufficient. The only other player with an incentive to provide money >> is the authors. My understanding is that the economics are forcing ACM >> to?go?in?that?direction. >> >> I believe ACM Is trying to be a good actor here, unlike publishers that >> double-dip by extracting money from both the authors (publication fees) >> and the readers (subscription fees); those publishers are doing very >> well financially and generating well-earned resentment. My understanding >> is that ACM does not want to double-dip. Instead, the idea is that >> authors at institutions with ACM subscriptions will pay lower or no fees >> for publications. That should keep the total cost to institutions under >> control and hopefully approximately cost-neutral. And note that the open >> access fees charged to other authors are still much lower than the >> author fees charged by other publishers. The journal Nature charges >> authors?$2000,?for?example,?and?it?is?not?the?high?end. >> >> Best, >> >> Andrew?Myers >> >> Gabriel?Scherer?wrote?on?12/21/19?6:01?AM: >>> [?The?Types?Forum, >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>> >>> Dear?Roberto?(and?list), >>> >>> The?new?ACM?Open?model?is?based?on?the?core?idea?of?saving?the?licensing >>> revenue?of?the?ACM?by?shifting?costs?from?their?many?customers?(including >>> in?particular?companies)?to?only?the?institutions?who?submit?the?articles. >>> >>> They?hope?that?the?academic?actors?that?produce?the?scientific?value?will >>> also?pay?for?current?ACM?expenses.?This?model?is?completely?incompatible >>> with?having?fair?Open?Access?prices?for?ACM?publications;?on?the?contrary, >>> >>> it?would?result?in?a?strong?total-cost?increase?for?academic?entities?that >>> >>> publish?in?ACM?proceedings. >>> >>> This?is?frankly?explained?on?the?(current?version?of)?the?ACM?Open >>> documentation?page: >>> https://www.acm.org/publications/openaccess#acmopen >>> >>>> Today,?ACM?Publications?and?the?ACM?Digital?Library?platform?are?funded?by >>>> >>>> selling?"read"?or?"access"?licenses?to?approximately?2,700?universities, >>>> government?research?labs,?and?corporations?from?around?the?world.?The >>>> income?generated?from?the?sale?of?these?licenses?[...]?is?approximately >>>> $20M+?annually >>>> >>> The?vast?majority?of?[ACM]?articles?are?authored?by?individuals?affiliated >>> >>>> with?~1,000?institutions,?which?is?roughly?1/3?of?the?institutions?that >>>> license??access??to?the?ACM?Digital?Library.?So,?the?main?challenge?for?ACM >>>> >>>> is?how?to?generate?roughly?the?same?income?from?1/3?the?number?of >>>> institutions?over?the?long?term,?as?ACM?transitions?from?selling >>>> institutional?"access"?to?an?institutional?"OA?publication"?model?and?more >>>> >>>> and?more?of?the?articles?published?in?the?ACM?DL?are?published?in?front?of >>>> >>>> the?subscription?paywall. >>>> >>> A?transition?to?fair?Open?Access?practices?would?require?the?difficult >>> decision?of?giving?up?on?licensing?revenue. >>> The?ACM?does?not?seem?willing?to?do?it,?and?cannot?be?trusted?to?do?it >>> eventually. >>> >>> >>> On?Fri,?Dec?20,?2019?at?7:08?PM?Roberto?Di?Cosmo >>> wrote: >>> >>>> Thanks?Gabriel?for?bringing?this?to?this?list:?it?was?indeed?shocking?to >>>> see?ACM >>>> (and?many?other?learned?societies)?in?the?list?of?signatories?of?this >>>> letter. >>>> >>>> The?fact?that?many?small?learned?societies?do?not?feel?ready?to?jump?into >>>> >>>> a?pure >>>> open?access?model?right?away?does?not?justify?their?signature?on?a?letter >>>> >>>> containing?highly?debatable?(that's?an?euphemism)?statements?like?the?ones >>>> >>>> you?pinpoint. >>>> >>>> By?a?curious?coincidence,?I?got?almost?at?the?same?time?an?ACM?newlsetter >>>> >>>> (Blue >>>> Diamond)?containing?among?other?announcements,?this?one: >>>> >>>> ?????ACM?OPEN:?A?New?Transformative?Model?for?Open?Access?Publication >>>> >>>> ??????Over?the?past?year?ACM?Publications?staff?have?been?working >>>> collaboratively?with >>>> ??????a?group?of?large?research?universities?in?the?United?States?to >>>> develop?an >>>> ??????entirely?new?and?innovative?model?for?Open?Access?publication?that >>>> has?the >>>> ??????potential?to?transition?ACM?into?a?predominantly?Open?Access >>>> publisher?over?the >>>> ??????next?decade?or?sooner. >>>> >>>> You?can?find?details?of?the?proposed?model?at >>>> https://www.acm.org/publications/openaccess#acmopen >>>> >>>> Cheers >>>> >>>> -- >>>> Roberto >>>> >>>> On?Fri,?Dec?20,?2019?at?02:53:05PM?+0100,?Gabriel?Scherer?wrote: >>>>> [?The?Types?Forum, >>>> http://lists.seas.upenn.edu/mailman/listinfo/types-list?] >>>>> Dear?types-list?and?SIGPLAN, >>>>> >>>>> I?have?long?been?of?the?opinion?that?our?scientific?publications >>>>> should?be?Open?Access,?and?that?editors?should?not?request?more?than >>>>> a?fair?price?(cost?of?publication,?which?Dasgtuhl?estimates?at?$60 >>>>> per?article).?In?particular,?I?believe?that?copyright?transfer >>>>> agreements,?as?imposed?by?most?editors?including?the?ACM,?is?deeply >>>>> unethical:?the?publishers?are?not?the?authors?of?our?scientific >>>>> production?and?they?should?not?force?us?to?give?our?copyright?to >>>>> them.?A?non-exclusive?publishing?agreement?should?be?enough. >>>>> >>>>> Whether?or?not?you?agree?with?this?position,?you?may?be?interested?in >>>>> the?content?of?the?following?letter?to?the?US?White?House?that >>>>> a?coalition?of?scientific?publishers,?*including?the?ACM*,?signed?and >>>>> support. >>>>> >>>>> >>>>> >>>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >>>> >>>>> ???press?release?from?the?coalition?of?editors: >>>>> >>>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >>>> >>>>> (This?letter?was?written?in?the?context?of?a?proposed?US?legislation >>>>> to?force?more?scientists?to?publish?their?work?in?Fair?Open?Access >>>>> venues.?I?haven't?been?able?to?find?a?precise?link?to?a?discussion?of >>>>> this?proposed?legislation.) >>>>> >>>>> The?following?parts?of?the?letter?co-signed?by?the?ACM?are >>>>> particularly?juicy: >>>>> >>>>>> [We]?have?learned?that?the?Administration?may?be?preparing?to?step >>>>>> into?the?private?marketplace?and?force?the?immediate?free?distribution >>>>>> of?journal?articles?financed?and?published?by?organizations?in?the >>>>>> private?sector,?including?many?non-profits.?This?would?effectively >>>>>> nationalize?the?valuable?American?intellectual?property?that?we >>>>>> produce?and?force?us?to?give?it?away?to?the?rest?of?the?world?for >>>>>> free. >>>>>> This?mandate?[...]?would?make?it?very?difficult?for?most?American >>>>>> publishers?to?invest?in?publishing?these?articles.?As?a?consequence, >>>>>> it?would?place?increased?financial?responsibility?on?the?government >>>>>> through?diverted?federal?research?grant?funds?or?additional?monies >>>>>> to?underwrite?the?important?value?added?by?publishing.?In?the?coming >>>>>> years,?this?cost?shift?would?place?billions?of?dollars?of?new?and >>>>>> additional?burden?on?taxpayers. >>>>> In?my?discussion?with?many?of?us,?I?regularly?hear?that?the?ACM?is >>>>> "not?evil"?(the?SIGPLAN,?of?course,?is?pure?good!)?and?that?placating >>>>> its?weird?views?(for?example,?that?it?really?does?cost?$700?or?$900?to >>>>> publish?an?article?as?Open?Source)?is?good?for?our?research >>>>> community.?It?do?not?see?how?this?argument?is?compatible?with?the?ACM >>>>> signing?this?letter. >>>>> >>>>> I?believe?that?many?of?our?activities,?which?we?collectively?trained >>>>> ourselves?to?see?as?harmless?administrative?details?of?our?research >>>>> work,?are?in?fact?empowering?the?ACM?to?make?those?claims.?Should?we >>>>> accept?to?give?away?our?copyright,?or?pay??unreasonable >>>>> Gold?Access?author?processing?charges?(APCs)? >>>>> >>>>> ?:?The?SIGPLAN?decision?to?cover?APC?costs?for?PACMPL?articles?is >>>>> shielding?many?of?us?from?paying?APCs.?But?many?of?the?smaller >>>>> conferences,?symposiums?or?workshops?in?our?community?whose >>>>> proceedings?are?handled?by?the?ACM?are?still?limited?to?"pay >>>>> $900"?(or?"pay?$25?per?page")?as?the?only?Open?Access?option,?with >>>>> copyright?transfer?as?the?only?free?choice,?which?is?effectively >>>>> keeping?those?proceedings?Closed-Access. >>>> -- >>>> Roberto?Di?Cosmo >>>> >>>> ------------------------------------------------------------------ >>>> Computer?Science?Professor >>>> ?????????????(on?leave?at?INRIA?from?IRIF/University?Paris?Diderot) >>>> >>>> Director >>>> Software?Heritage https://www.softwareheritage.org >>>> INRIA >>>> Bureau?C328??????????????????E-mail?: roberto at dicosmo.org >>>> 2,?Rue?Simone?Iff??????????Web?page?: http://www.dicosmo.org >>>> CS?42112????????????????????Twitter?: http://twitter.com/rdicosmo >>>> 75589?Paris?Cedex?12????????????Tel?:?+33?1?80?49?44?42 >>>> ------------------------------------------------------------------ >>>> >>>> GPG?fingerprint?2931?20CE?3A5A?5390?98EC?8BFC?FCCA?C3BE?39CB?12D3 >>>> >>>> >> >> >> ------------------------------ >> >> Message: 2 >> Date: Sun, 22 Dec 2019 18:52:12 +0100 >> From: Gabriel Scherer >> To: Andrew Myers >> Cc: Types list >> Subject: Re: [TYPES] In a letter to the US White House, ACM opposes >> free distribution of peer-reviewed journal articles >> Message-ID: >> >> Content-Type: text/plain; charset="UTF-8" >> >> Dear Andrew (and list), >> >> I believe open access is a goal for ACM >> >> >> This is what the ACM says, but this is not their actions suggest. Some >> examples: >> >> 1. They signed this letter. (They defend their choice in >> https://www.acm.org/about-acm/opposition-to-zero-embargo-mandate ) >> >> 2. Events affiliated with an ACM conference, such as a workshop, are not >> allowed to publish their proceedings as (fair) open-access if they wish to, >> for example by publishing in ETCS or LiPICS. (I know from my experience >> with the ML and OCaml workshops that ACM people check this and enforce this >> rule.) >> >> 3. According to private communication with ETAPS organizers, the Gold Open >> Access deal offered by Springer costs *less* per paper for ETAPS than the >> Open Access model that SIGPLAN generously funds for PACMPL. If you're doing >> worse than Springer at Open Access, you are probably not trying very hard. >> >> I hope we can agree that publishers do provide some value in supporting the >>> scientific process, for example by maintaining archives of publications for >>> decades and across formats. >>> >> According to LiPICS (the fair Open Access publishing arm of Schloss >> Dagstuhl), their edition/typesetting work costs 60? per article ( >> https://www.dagstuhl.de/en/publications/lipics/processing-charge/ ). (In >> any case, ACM outsources their edition work on proceedings to external >> companies, that if I understand correctly are budgeted as part of the >> conference organization, so not paid by ACM itself.) >> >> According to arXiv, their long-term archival platform costs <$7 per article >> ( https://arxiv.org/help/support/whitepaper#21-budget ). >> >> On Sun, Dec 22, 2019 at 3:46 PM Andrew Myers wrote: >> >>> It feels a bit facile to bash the ACM for signing onto this letter. The >>> letter does not mean that they oppose making publications freely available; >>> in fact, I believe open access is a goal for ACM. The letter means that >>> they oppose having the government *mandate* that all scientific publishers >>> operate in this way. Exactly what the right funding model is for scientific >>> publications is still up in the air. Should the government spend taxes >>> enforcing rules whose implications we do not fully understand? I think not. >>> >>> The discussions I have seen about this topic seem to focus on the costs to >>> readers and authors while completely ignoring the economics of publishing. >>> I hope we can agree that publishers do provide some value in supporting the >>> scientific process, for example by maintaining archives of publications for >>> decades and across formats. That value can only be delivered if ACM et al. >>> have money. Where are they supposed to get it? The old model of libraries >>> paying ACM subscriptions is dying and is incompatible with open access. >>> Corporate charity is unreliable and insufficient. The only other player >>> with an incentive to provide money is the authors. My understanding is that >>> the economics are forcing ACM to go in that direction. >>> >>> I believe ACM Is trying to be a good actor here, unlike publishers that >>> double-dip by extracting money from both the authors (publication fees) and >>> the readers (subscription fees); those publishers are doing very well >>> financially and generating well-earned resentment. My understanding is that >>> ACM does not want to double-dip. Instead, the idea is that authors at >>> institutions with ACM subscriptions will pay lower or no fees for >>> publications. That should keep the total cost to institutions under control >>> and hopefully approximately cost-neutral. And note that the open access >>> fees charged to other authors are still much lower than the author fees >>> charged by other publishers. The journal Nature charges >>> authors $2000, for example, and it is not the high end. >>> >>> Best, >>> >>> Andrew Myers >>> >>> Gabriel Scherer wrote on 12/21/19 6:01 AM: >>> >>> [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >>> >>> >>> Dear Roberto (and list), >>> >>> The new ACM Open model is based on the core idea of saving the licensing >>> revenue of the ACM by shifting costs from their many customers (including >>> in particular companies) to only the institutions who submit the articles. >>> They hope that the academic actors that produce the scientific value will >>> also pay for current ACM expenses. This model is completely incompatible >>> with having fair Open Access prices for ACM publications; on the contrary, >>> it would result in a strong total-cost increase for academic entities that >>> publish in ACM proceedings. >>> >>> This is frankly explained on the (current version of) the ACM Open >>> documentation page: >>> https://www.acm.org/publications/openaccess#acmopen >>> >>> Today, ACM Publications and the ACM Digital Library platform are funded by >>> selling "read" or "access" licenses to approximately 2,700 universities, >>> government research labs, and corporations from around the world. The >>> income generated from the sale of these licenses [...] is approximately >>> $20M+ annually >>> >>> The vast majority of [ACM] articles are authored by individuals affiliated >>> >>> with ~1,000 institutions, which is roughly 1/3 of the institutions that >>> license ?access? to the ACM Digital Library. So, the main challenge for ACM >>> >>> is how to generate roughly the same income from 1/3 the number of >>> institutions over the long term, as ACM transitions from selling >>> institutional "access" to an institutional "OA publication" model and more >>> and more of the articles published in the ACM DL are published in front of >>> the subscription paywall. >>> >>> A transition to fair Open Access practices would require the difficult >>> decision of giving up on licensing revenue. >>> The ACM does not seem willing to do it, and cannot be trusted to do it >>> eventually. >>> >>> >>> On Fri, Dec 20, 2019 at 7:08 PM Roberto Di Cosmo >>> >>> wrote: >>> >>> Thanks Gabriel for bringing this to this list: it was indeed shocking to >>> see ACM >>> (and many other learned societies) in the list of signatories of this >>> letter. >>> >>> The fact that many small learned societies do not feel ready to jump into >>> a pure >>> open access model right away does not justify their signature on a letter >>> containing highly debatable (that's an euphemism) statements like the ones >>> you pinpoint. >>> >>> By a curious coincidence, I got almost at the same time an ACM newlsetter >>> (Blue >>> Diamond) containing among other announcements, this one: >>> >>> ACM OPEN: A New Transformative Model for Open Access Publication >>> >>> Over the past year ACM Publications staff have been working >>> collaboratively with >>> a group of large research universities in the United States to >>> develop an >>> entirely new and innovative model for Open Access publication that >>> has the >>> potential to transition ACM into a predominantly Open Access >>> publisher over the >>> next decade or sooner. >>> >>> You can find details of the proposed model at >>> https://www.acm.org/publications/openaccess#acmopen >>> >>> Cheers >>> >>> -- >>> Roberto >>> >>> On Fri, Dec 20, 2019 at 02:53:05PM +0100, Gabriel Scherer wrote: >>> >>> [ The Types Forum, >>> >>> http://lists.seas.upenn.edu/mailman/listinfo/types-list ] >>> >>> Dear types-list and SIGPLAN, >>> >>> I have long been of the opinion that our scientific publications >>> should be Open Access, and that editors should not request more than >>> a fair price (cost of publication, which Dasgtuhl estimates at $60 >>> per article). In particular, I believe that copyright transfer >>> agreements, as imposed by most editors including the ACM, is deeply >>> unethical: the publishers are not the authors of our scientific >>> production and they should not force us to give our copyright to >>> them. A non-exclusive publishing agreement should be enough. >>> >>> Whether or not you agree with this position, you may be interested in >>> the content of the following letter to the US White House that >>> a coalition of scientific publishers, *including the ACM*, signed and >>> support. >>> >>> >>> >>> >>> https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf >>> >>> press release from the coalition of editors: >>> >>> >>> https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles >>> >>> (This letter was written in the context of a proposed US legislation >>> to force more scientists to publish their work in Fair Open Access >>> venues. I haven't been able to find a precise link to a discussion of >>> this proposed legislation.) >>> >>> The following parts of the letter co-signed by the ACM are >>> particularly juicy: >>> >>> [We] have learned that the Administration may be preparing to step >>> into the private marketplace and force the immediate free distribution >>> of journal articles financed and published by organizations in the >>> private sector, including many non-profits. This would effectively >>> nationalize the valuable American intellectual property that we >>> produce and force us to give it away to the rest of the world for >>> free. >>> This mandate [...] would make it very difficult for most American >>> publishers to invest in publishing these articles. As a consequence, >>> it would place increased financial responsibility on the government >>> through diverted federal research grant funds or additional monies >>> to underwrite the important value added by publishing. In the coming >>> years, this cost shift would place billions of dollars of new and >>> additional burden on taxpayers. >>> >>> In my discussion with many of us, I regularly hear that the ACM is >>> "not evil" (the SIGPLAN, of course, is pure good!) and that placating >>> its weird views (for example, that it really does cost $700 or $900 to >>> publish an article as Open Source) is good for our research >>> community. It do not see how this argument is compatible with the ACM >>> signing this letter. >>> >>> I believe that many of our activities, which we collectively trained >>> ourselves to see as harmless administrative details of our research >>> work, are in fact empowering the ACM to make those claims. Should we >>> accept to give away our copyright, or pay? unreasonable >>> Gold Access author processing charges (APCs)? >>> >>> ?: The SIGPLAN decision to cover APC costs for PACMPL articles is >>> shielding many of us from paying APCs. But many of the smaller >>> conferences, symposiums or workshops in our community whose >>> proceedings are handled by the ACM are still limited to "pay >>> $900" (or "pay $25 per page") as the only Open Access option, with >>> copyright transfer as the only free choice, which is effectively >>> keeping those proceedings Closed-Access. >>> >>> -- >>> Roberto Di Cosmo >>> >>> ------------------------------------------------------------------ >>> Computer Science Professor >>> (on leave at INRIA from IRIF/University Paris Diderot) >>> >>> Director >>> Software Heritage https://www.softwareheritage.org >>> INRIA >>> Bureau C328 E-mail : roberto at dicosmo.org >>> 2, Rue Simone Iff Web page : http://www.dicosmo.org >>> CS 42112 Twitter : http://twitter.com/rdicosmo >>> 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 >>> ------------------------------------------------------------------ >>> >>> GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 >>> >>> >>> >>> >> >> ------------------------------ >> >> Subject: Digest Footer >> >> _______________________________________________ >> Types-list mailing list >> Types-list at LISTS.SEAS.UPENN.EDU >> https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list >> >> >> ------------------------------ >> >> End of Types-list Digest, Vol 121, Issue 4 >> ****************************************** From valeria.depaiva at gmail.com Wed Dec 25 09:58:09 2019 From: valeria.depaiva at gmail.com (Valeria de Paiva) Date: Wed, 25 Dec 2019 06:58:09 -0800 Subject: [TYPES] Types-list Digest, Vol 121, Issue 8 In-Reply-To: References: Message-ID: dear all, I'd like to add my 2 cents to this thread, replying to this bit of Roberto's message. >On the one hand, it created a gigantic legacy problem: even if we all started publishing Open Access today, what about the hundreds of thousands of articles published over the past decades, that are still behind a paywall? II would like to volunteer the information that the ACL the Association for Computational Linguistics https://www.aclweb.org/anthology/ did exactly this. It made all the articles in their main conferences free and open access; and then they went back and digitalized and open-sourced their past! they made the already published papers also open source. Now if the computational linguists can do it, I fail to see why we cannot. To me, it seems a lack of will from our learned societies. But I don't know the numbers, of course. Happy holidays! Valeria On Wed, Dec 25, 2019 at 1:28 AM wrote: > Send Types-list mailing list submissions to > types-list at LISTS.SEAS.UPENN.EDU > > To subscribe or unsubscribe via the World Wide Web, visit > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > or, via email, send a message with subject or body 'help' to > types-list-request at LISTS.SEAS.UPENN.EDU > > You can reach the person managing the list at > types-list-owner at LISTS.SEAS.UPENN.EDU > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Types-list digest..." > > > Today's Topics: > > 1. Re: In a letter to the US White House, (Robert Rand) > 2. Re: In a letter to the US White House, (Gabriel Scherer) > 3. Open Access: a bit of background (Was: Re: In a letter to > the US White House, ACM opposes free distribution of > peer-reviewed journal articles) (Roberto Di Cosmo) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Sun, 22 Dec 2019 21:11:42 -0500 > From: Robert Rand > To: Peter Selinger > Cc: Types list > Subject: Re: [TYPES] In a letter to the US White House, > Message-ID: > < > CAGKXo9deS4hC+Nj_anQ9de3nZ4KfH53Mg6bAPpOhhvgKtadHkw at mail.gmail.com> > Content-Type: text/plain; charset="UTF-8" > > I agree with Peter and Gabriel. The ACM seems intent on charging the same > amount per article and just shifting around who pays for it: from readers > to authors to conference-goers to (in their newest scheme) universities as > part of some sort of general tax. And we shouldn't be paying it. > > I would add that for publishing conference proceedings, EPTCS (which > publishes the conference proceedings for Linearity, ICLP, QPL and dozens of > other conferences) is a great option. EPTCS is free to everyone, and it > would be nice if more programming languages conferences switched to using > it. > > On Sun, Dec 22, 2019 at 12:38 PM wrote: > > > [ The Types Forum, > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > ] > > > > I was recently invited to the editorial board of a new ACM journal. I > > refused to become an editor, on the grounds that the journal is not > > open access. > > > > I feel that as editors, we volunteer our time, and I have decided to > > do so only for journals that are truly open access. And by "open > > access", I do not mean some "gold" model based on author charges > > (currently $1300-$1700 per ACM journal article, see > > https://www.acm.org/publications/openaccess). I mean journals that are > > free for both authors and readers. > > > > There are already many such journals, and they are usually > > community-run. Examples include Logical Methods in Computer Science > > (https://lmcs.episciences.org/), Compositionality > > (https://www.compositionality-journal.org/), and Quantum > > (https://quantum-journal.org/). There is relatively little in the way > > of interested groups of individuals starting new such journals. > > > > I think the best way to get rid of unreasonable publishing models is > > not to participate in them. > > > > -- Peter > > > > Jonathan Aldrich wrote: > > > > > > > > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > > gabriel.scherer at gmail.com> > > > wrote: > > > > > > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > > ] > > > > > > > > Dear types-list and SIGPLAN, > > > > > > > > I have long been of the opinion that our scientific publications > > > > should be Open Access, and that editors should not request more than > > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > > per article). In particular, I believe that copyright transfer > > > > agreements, as imposed by most editors including the ACM, is deeply > > > > unethical: the publishers are not the authors of our scientific > > > > production and they should not force us to give our copyright to > > > > them. A non-exclusive publishing agreement should be enough. > > > > > > > > Whether or not you agree with this position, you may be interested in > > > > the content of the following letter to the US White House that > > > > a coalition of scientific publishers, *including the ACM*, signed and > > > > support. > > > > > > > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > > press release from the coalition of editors: > > > > > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > > > (This letter was written in the context of a proposed US legislation > > > > to force more scientists to publish their work in Fair Open Access > > > > venues. I haven't been able to find a precise link to a discussion of > > > > this proposed legislation.) > > > > > > > > The following parts of the letter co-signed by the ACM are > > > > particularly juicy: > > > > > > > > > [We] have learned that the Administration may be preparing to step > > > > > into the private marketplace and force the immediate free > > distribution > > > > > of journal articles financed and published by organizations in the > > > > > private sector, including many non-profits. This would effectively > > > > > nationalize the valuable American intellectual property that we > > > > > produce and force us to give it away to the rest of the world for > > > > > free. > > > > > > > > > This mandate [...] would make it very difficult for most American > > > > > publishers to invest in publishing these articles. As a > consequence, > > > > > it would place increased financial responsibility on the government > > > > > through diverted federal research grant funds or additional monies > > > > > to underwrite the important value added by publishing. In the > coming > > > > > years, this cost shift would place billions of dollars of new and > > > > > additional burden on taxpayers. > > > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > > its weird views (for example, that it really does cost $700 or $900 > to > > > > publish an article as Open Source) is good for our research > > > > community. It do not see how this argument is compatible with the ACM > > > > signing this letter. > > > > > > > > I believe that many of our activities, which we collectively trained > > > > ourselves to see as harmless administrative details of our research > > > > work, are in fact empowering the ACM to make those claims. Should we > > > > accept to give away our copyright, or pay? unreasonable > > > > Gold Access author processing charges (APCs)? > > > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > > shielding many of us from paying APCs. But many of the smaller > > > > conferences, symposiums or workshops in our community whose > > > > proceedings are handled by the ACM are still limited to "pay > > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > > copyright transfer as the only free choice, which is effectively > > > > keeping those proceedings Closed-Access. > > > > > > > > > > > > > > > > > ------------------------------ > > Message: 2 > Date: Mon, 23 Dec 2019 16:28:35 +0100 > From: Gabriel Scherer > To: Peter Selinger > Cc: Jonathan Aldrich , Types list > > Subject: Re: [TYPES] In a letter to the US White House, > Message-ID: > EOhoj9w1tUWMOHBrFSEfnoT4ri6L+HtmhtymUMww at mail.gmail.com> > Content-Type: text/plain; charset="UTF-8" > > I strongly agree with Peter: the most effective action against unreasonable > publishing models is to not participate in them. We are very fortunate to > work in a community where many sensible Open Access venues exist, so that > this decision comes at little to no personal cost. > > I created a webpage to document this policy (tentatively named "Keeping my > work Open"). To evaluate the impact on my work since I started applying it > in 2017, I listed venues to which I did or did not participate due to the > policy. > > http://gasche.info/open_access.html > > The summary is that this principle is not limiting: most venues in > Programming Languages research have sensible policies (I include PACMPL, > thanks to the generous support of SIGPLAN to fund APCs, and ETAPS since > 2018), and that the Closed Access venues I encountered had fairly direct > replacements in all cases. > > On Sun, Dec 22, 2019 at 7:37 AM wrote: > > > I was recently invited to the editorial board of a new ACM journal. I > > refused to become an editor, on the grounds that the journal is not > > open access. > > > > I feel that as editors, we volunteer our time, and I have decided to > > do so only for journals that are truly open access. And by "open > > access", I do not mean some "gold" model based on author charges > > (currently $1300-$1700 per ACM journal article, see > > https://www.acm.org/publications/openaccess). I mean journals that are > > free for both authors and readers. > > > > There are already many such journals, and they are usually > > community-run. Examples include Logical Methods in Computer Science > > (https://lmcs.episciences.org/), Compositionality > > (https://www.compositionality-journal.org/), and Quantum > > (https://quantum-journal.org/). There is relatively little in the way > > of interested groups of individuals starting new such journals. > > > > I think the best way to get rid of unreasonable publishing models is > > not to participate in them. > > > > -- Peter > > > > Jonathan Aldrich wrote: > > > > > > > > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > > gabriel.scherer at gmail.com> > > > wrote: > > > > > > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > > ] > > > > > > > > Dear types-list and SIGPLAN, > > > > > > > > I have long been of the opinion that our scientific publications > > > > should be Open Access, and that editors should not request more than > > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > > per article). In particular, I believe that copyright transfer > > > > agreements, as imposed by most editors including the ACM, is deeply > > > > unethical: the publishers are not the authors of our scientific > > > > production and they should not force us to give our copyright to > > > > them. A non-exclusive publishing agreement should be enough. > > > > > > > > Whether or not you agree with this position, you may be interested in > > > > the content of the following letter to the US White House that > > > > a coalition of scientific publishers, *including the ACM*, signed and > > > > support. > > > > > > > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > > press release from the coalition of editors: > > > > > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > > > (This letter was written in the context of a proposed US legislation > > > > to force more scientists to publish their work in Fair Open Access > > > > venues. I haven't been able to find a precise link to a discussion of > > > > this proposed legislation.) > > > > > > > > The following parts of the letter co-signed by the ACM are > > > > particularly juicy: > > > > > > > > > [We] have learned that the Administration may be preparing to step > > > > > into the private marketplace and force the immediate free > > distribution > > > > > of journal articles financed and published by organizations in the > > > > > private sector, including many non-profits. This would effectively > > > > > nationalize the valuable American intellectual property that we > > > > > produce and force us to give it away to the rest of the world for > > > > > free. > > > > > > > > > This mandate [...] would make it very difficult for most American > > > > > publishers to invest in publishing these articles. As a > consequence, > > > > > it would place increased financial responsibility on the government > > > > > through diverted federal research grant funds or additional monies > > > > > to underwrite the important value added by publishing. In the > coming > > > > > years, this cost shift would place billions of dollars of new and > > > > > additional burden on taxpayers. > > > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > > its weird views (for example, that it really does cost $700 or $900 > to > > > > publish an article as Open Source) is good for our research > > > > community. It do not see how this argument is compatible with the ACM > > > > signing this letter. > > > > > > > > I believe that many of our activities, which we collectively trained > > > > ourselves to see as harmless administrative details of our research > > > > work, are in fact empowering the ACM to make those claims. Should we > > > > accept to give away our copyright, or pay? unreasonable > > > > Gold Access author processing charges (APCs)? > > > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > > shielding many of us from paying APCs. But many of the smaller > > > > conferences, symposiums or workshops in our community whose > > > > proceedings are handled by the ACM are still limited to "pay > > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > > copyright transfer as the only free choice, which is effectively > > > > keeping those proceedings Closed-Access. > > > > > > > > > > > > > > > > > ------------------------------ > > Message: 3 > Date: Tue, 24 Dec 2019 18:21:38 +0100 > From: Roberto Di Cosmo > To: Gabriel Scherer , Types list > > Subject: [TYPES] Open Access: a bit of background (Was: Re: In a > letter to the US White House, ACM opposes free distribution of > peer-reviewed journal articles) > Message-ID: <20191224172138.GC32219 at traveler> > Content-Type: text/plain; charset=us-ascii > > Dear all, > I'd like to contribute to this very interesting thread by offering > some > background on the Open Acces debate, that started almost a quarter of a > century > ago, and that may help younger collegues to get an idea of why we ended up > where we are today. > > It's important to recall that, no matter the official statements one can > read > here and there, the goals of researchers on one side, and traditional > scientific > publishers on the other, *have always been quite different*: researchers > wanted > access to publications as broad and efficient as possible, and a means to > signal > excellence of their work; publishers wanted to run a healthy business, and > a > monopoly on their own areas (via mandatory copyright transfer and > ownership of > the publication trademarks). > > It just turns out that in the pre-Internet era what publishers offered also > satisfied the goals of researchers: the difficult divorce started some > twenty > years ago, with the generalisation of the Internet, when sharing papers on > the > Web became a *more efficient* means of distibution than having copies of > journals or proceedings sent to thousands of libraries all over the world, > and > the publishers' own goals started to become an obstacle to the free > dissemination of science. > > Unfortunately, before this divorce started, northern emisphere researchers > (*), as > well as funding agencies, had broadly agreed to the conditions imposed by > publishers, in particular mandatory transfer to them of exclusive > copyright, > which were not necessary to achieve the researcher's goals. > > This had quite disastrous consequences. > > On the one hand, it created a gigantic legacy problem: even if we all > started > publishing Open Access today, what about the hunderds of thousands of > articles > published over the past decades, that are still behind a paywall? > > On the other hand, and maybe even worse, it accustomed all the players > (including governments and funding agencies) to accept the idea that the > publishing houses actually *own* the copyright to the articles we write, > while > they actually *extort* this copyright from the authors themselves, by > forcing > them to relinquish their rights in order to be published, a provision that > is in > direct violation of the spirit of copyright itself (in France, it is > actually in > violation of the letter of copyright law, see article L.131-4 of the CPI). > > You can find a trace of this line of thinking in the wording used in the > letter > that started this whole thread, where it mentions "intellectual property" > that belongs to the publishers. > > A broader and longer analysis of what is at stake can be found in this > early > account I wrote 15 years ago, when some of us had the illusion that Open > Access > was going to win soon : > > Scientific Publications: The Role of Public Administrations in The ICT > Era, Upgrade, 2006 > + available at: http://www.dicosmo.org/FSP/FATOS-Upgrade-03-2006.pdf > + french original version available at > http://www.dicosmo.org/FSP/FreeAccessToScience.pdf > > Cheers > > -- > Roberto > > (*) in Latin America the situation is quite different > > ------------------------------------------------------------------ > Computer Science Professor > (on leave at INRIA from IRIF/University Paris Diderot) > > Director > Software Heritage https://www.softwareheritage.org > INRIA > Bureau C328 E-mail : roberto at dicosmo.org > 2, Rue Simone Iff Web page : http://www.dicosmo.org > CS 42112 Twitter : http://twitter.com/rdicosmo > 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 > ------------------------------------------------------------------ > > GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 > > > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > Types-list mailing list > Types-list at LISTS.SEAS.UPENN.EDU > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > > > ------------------------------ > > End of Types-list Digest, Vol 121, Issue 8 > ****************************************** > -- Valeria de Paiva http://vcvpaiva.github.io/ http://www.cs.bham.ac.uk/~vdp/ From marco.servetto at gmail.com Wed Dec 25 15:09:20 2019 From: marco.servetto at gmail.com (Marco Servetto) Date: Thu, 26 Dec 2019 09:09:20 +1300 Subject: [TYPES] Types-list Digest, Vol 121, Issue 8 In-Reply-To: References: Message-ID: This seams to be a point against the "the best way is to just not participate and choose open access alternatives" point of view. By directly pressuring ACM and similar we could improve the situation on the "old" documents too. On Thu, 26 Dec 2019 at 05:52, Valeria de Paiva wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > dear all, > I'd like to add my 2 cents to this thread, replying to this bit of > Roberto's message. > > >On the one hand, it created a gigantic legacy problem: even if we all > started > publishing Open Access today, what about the hundreds of thousands of > articles > published over the past decades, that are still behind a paywall? > > II would like to volunteer the information that the ACL the Association for > Computational Linguistics > https://www.aclweb.org/anthology/ > did exactly this. It made all the articles in their main conferences free > and open access; and then they went back and > digitalized and open-sourced their past! they made the already published > papers also open source. > Now if the computational linguists can do it, I fail to see why we cannot. > To me, it seems a lack of will from our learned societies. But I don't know > the numbers, of course. > > Happy holidays! > Valeria > > On Wed, Dec 25, 2019 at 1:28 AM > wrote: > > > Send Types-list mailing list submissions to > > types-list at LISTS.SEAS.UPENN.EDU > > > > To subscribe or unsubscribe via the World Wide Web, visit > > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > > or, via email, send a message with subject or body 'help' to > > types-list-request at LISTS.SEAS.UPENN.EDU > > > > You can reach the person managing the list at > > types-list-owner at LISTS.SEAS.UPENN.EDU > > > > When replying, please edit your Subject line so it is more specific > > than "Re: Contents of Types-list digest..." > > > > > > Today's Topics: > > > > 1. Re: In a letter to the US White House, (Robert Rand) > > 2. Re: In a letter to the US White House, (Gabriel Scherer) > > 3. Open Access: a bit of background (Was: Re: In a letter to > > the US White House, ACM opposes free distribution of > > peer-reviewed journal articles) (Roberto Di Cosmo) > > > > > > ---------------------------------------------------------------------- > > > > Message: 1 > > Date: Sun, 22 Dec 2019 21:11:42 -0500 > > From: Robert Rand > > To: Peter Selinger > > Cc: Types list > > Subject: Re: [TYPES] In a letter to the US White House, > > Message-ID: > > < > > CAGKXo9deS4hC+Nj_anQ9de3nZ4KfH53Mg6bAPpOhhvgKtadHkw at mail.gmail.com> > > Content-Type: text/plain; charset="UTF-8" > > > > I agree with Peter and Gabriel. The ACM seems intent on charging the same > > amount per article and just shifting around who pays for it: from readers > > to authors to conference-goers to (in their newest scheme) universities as > > part of some sort of general tax. And we shouldn't be paying it. > > > > I would add that for publishing conference proceedings, EPTCS (which > > publishes the conference proceedings for Linearity, ICLP, QPL and dozens of > > other conferences) is a great option. EPTCS is free to everyone, and it > > would be nice if more programming languages conferences switched to using > > it. > > > > On Sun, Dec 22, 2019 at 12:38 PM wrote: > > > > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > ] > > > > > > I was recently invited to the editorial board of a new ACM journal. I > > > refused to become an editor, on the grounds that the journal is not > > > open access. > > > > > > I feel that as editors, we volunteer our time, and I have decided to > > > do so only for journals that are truly open access. And by "open > > > access", I do not mean some "gold" model based on author charges > > > (currently $1300-$1700 per ACM journal article, see > > > https://www.acm.org/publications/openaccess). I mean journals that are > > > free for both authors and readers. > > > > > > There are already many such journals, and they are usually > > > community-run. Examples include Logical Methods in Computer Science > > > (https://lmcs.episciences.org/), Compositionality > > > (https://www.compositionality-journal.org/), and Quantum > > > (https://quantum-journal.org/). There is relatively little in the way > > > of interested groups of individuals starting new such journals. > > > > > > I think the best way to get rid of unreasonable publishing models is > > > not to participate in them. > > > > > > -- Peter > > > > > > Jonathan Aldrich wrote: > > > > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > > > gabriel.scherer at gmail.com> > > > > wrote: > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > > > ] > > > > > > > > > > Dear types-list and SIGPLAN, > > > > > > > > > > I have long been of the opinion that our scientific publications > > > > > should be Open Access, and that editors should not request more than > > > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > > > per article). In particular, I believe that copyright transfer > > > > > agreements, as imposed by most editors including the ACM, is deeply > > > > > unethical: the publishers are not the authors of our scientific > > > > > production and they should not force us to give our copyright to > > > > > them. A non-exclusive publishing agreement should be enough. > > > > > > > > > > Whether or not you agree with this position, you may be interested in > > > > > the content of the following letter to the US White House that > > > > > a coalition of scientific publishers, *including the ACM*, signed and > > > > > support. > > > > > > > > > > > > > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > > > press release from the coalition of editors: > > > > > > > > > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > > > > > (This letter was written in the context of a proposed US legislation > > > > > to force more scientists to publish their work in Fair Open Access > > > > > venues. I haven't been able to find a precise link to a discussion of > > > > > this proposed legislation.) > > > > > > > > > > The following parts of the letter co-signed by the ACM are > > > > > particularly juicy: > > > > > > > > > > > [We] have learned that the Administration may be preparing to step > > > > > > into the private marketplace and force the immediate free > > > distribution > > > > > > of journal articles financed and published by organizations in the > > > > > > private sector, including many non-profits. This would effectively > > > > > > nationalize the valuable American intellectual property that we > > > > > > produce and force us to give it away to the rest of the world for > > > > > > free. > > > > > > > > > > > This mandate [...] would make it very difficult for most American > > > > > > publishers to invest in publishing these articles. As a > > consequence, > > > > > > it would place increased financial responsibility on the government > > > > > > through diverted federal research grant funds or additional monies > > > > > > to underwrite the important value added by publishing. In the > > coming > > > > > > years, this cost shift would place billions of dollars of new and > > > > > > additional burden on taxpayers. > > > > > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > > > its weird views (for example, that it really does cost $700 or $900 > > to > > > > > publish an article as Open Source) is good for our research > > > > > community. It do not see how this argument is compatible with the ACM > > > > > signing this letter. > > > > > > > > > > I believe that many of our activities, which we collectively trained > > > > > ourselves to see as harmless administrative details of our research > > > > > work, are in fact empowering the ACM to make those claims. Should we > > > > > accept to give away our copyright, or pay? unreasonable > > > > > Gold Access author processing charges (APCs)? > > > > > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > > > shielding many of us from paying APCs. But many of the smaller > > > > > conferences, symposiums or workshops in our community whose > > > > > proceedings are handled by the ACM are still limited to "pay > > > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > > > copyright transfer as the only free choice, which is effectively > > > > > keeping those proceedings Closed-Access. > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------ > > > > Message: 2 > > Date: Mon, 23 Dec 2019 16:28:35 +0100 > > From: Gabriel Scherer > > To: Peter Selinger > > Cc: Jonathan Aldrich , Types list > > > > Subject: Re: [TYPES] In a letter to the US White House, > > Message-ID: > > > EOhoj9w1tUWMOHBrFSEfnoT4ri6L+HtmhtymUMww at mail.gmail.com> > > Content-Type: text/plain; charset="UTF-8" > > > > I strongly agree with Peter: the most effective action against unreasonable > > publishing models is to not participate in them. We are very fortunate to > > work in a community where many sensible Open Access venues exist, so that > > this decision comes at little to no personal cost. > > > > I created a webpage to document this policy (tentatively named "Keeping my > > work Open"). To evaluate the impact on my work since I started applying it > > in 2017, I listed venues to which I did or did not participate due to the > > policy. > > > > http://gasche.info/open_access.html > > > > The summary is that this principle is not limiting: most venues in > > Programming Languages research have sensible policies (I include PACMPL, > > thanks to the generous support of SIGPLAN to fund APCs, and ETAPS since > > 2018), and that the Closed Access venues I encountered had fairly direct > > replacements in all cases. > > > > On Sun, Dec 22, 2019 at 7:37 AM wrote: > > > > > I was recently invited to the editorial board of a new ACM journal. I > > > refused to become an editor, on the grounds that the journal is not > > > open access. > > > > > > I feel that as editors, we volunteer our time, and I have decided to > > > do so only for journals that are truly open access. And by "open > > > access", I do not mean some "gold" model based on author charges > > > (currently $1300-$1700 per ACM journal article, see > > > https://www.acm.org/publications/openaccess). I mean journals that are > > > free for both authors and readers. > > > > > > There are already many such journals, and they are usually > > > community-run. Examples include Logical Methods in Computer Science > > > (https://lmcs.episciences.org/), Compositionality > > > (https://www.compositionality-journal.org/), and Quantum > > > (https://quantum-journal.org/). There is relatively little in the way > > > of interested groups of individuals starting new such journals. > > > > > > I think the best way to get rid of unreasonable publishing models is > > > not to participate in them. > > > > > > -- Peter > > > > > > Jonathan Aldrich wrote: > > > > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > > > gabriel.scherer at gmail.com> > > > > wrote: > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > > > ] > > > > > > > > > > Dear types-list and SIGPLAN, > > > > > > > > > > I have long been of the opinion that our scientific publications > > > > > should be Open Access, and that editors should not request more than > > > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > > > per article). In particular, I believe that copyright transfer > > > > > agreements, as imposed by most editors including the ACM, is deeply > > > > > unethical: the publishers are not the authors of our scientific > > > > > production and they should not force us to give our copyright to > > > > > them. A non-exclusive publishing agreement should be enough. > > > > > > > > > > Whether or not you agree with this position, you may be interested in > > > > > the content of the following letter to the US White House that > > > > > a coalition of scientific publishers, *including the ACM*, signed and > > > > > support. > > > > > > > > > > > > > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > > > press release from the coalition of editors: > > > > > > > > > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > > > > > (This letter was written in the context of a proposed US legislation > > > > > to force more scientists to publish their work in Fair Open Access > > > > > venues. I haven't been able to find a precise link to a discussion of > > > > > this proposed legislation.) > > > > > > > > > > The following parts of the letter co-signed by the ACM are > > > > > particularly juicy: > > > > > > > > > > > [We] have learned that the Administration may be preparing to step > > > > > > into the private marketplace and force the immediate free > > > distribution > > > > > > of journal articles financed and published by organizations in the > > > > > > private sector, including many non-profits. This would effectively > > > > > > nationalize the valuable American intellectual property that we > > > > > > produce and force us to give it away to the rest of the world for > > > > > > free. > > > > > > > > > > > This mandate [...] would make it very difficult for most American > > > > > > publishers to invest in publishing these articles. As a > > consequence, > > > > > > it would place increased financial responsibility on the government > > > > > > through diverted federal research grant funds or additional monies > > > > > > to underwrite the important value added by publishing. In the > > coming > > > > > > years, this cost shift would place billions of dollars of new and > > > > > > additional burden on taxpayers. > > > > > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > > > its weird views (for example, that it really does cost $700 or $900 > > to > > > > > publish an article as Open Source) is good for our research > > > > > community. It do not see how this argument is compatible with the ACM > > > > > signing this letter. > > > > > > > > > > I believe that many of our activities, which we collectively trained > > > > > ourselves to see as harmless administrative details of our research > > > > > work, are in fact empowering the ACM to make those claims. Should we > > > > > accept to give away our copyright, or pay? unreasonable > > > > > Gold Access author processing charges (APCs)? > > > > > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > > > shielding many of us from paying APCs. But many of the smaller > > > > > conferences, symposiums or workshops in our community whose > > > > > proceedings are handled by the ACM are still limited to "pay > > > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > > > copyright transfer as the only free choice, which is effectively > > > > > keeping those proceedings Closed-Access. > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------ > > > > Message: 3 > > Date: Tue, 24 Dec 2019 18:21:38 +0100 > > From: Roberto Di Cosmo > > To: Gabriel Scherer , Types list > > > > Subject: [TYPES] Open Access: a bit of background (Was: Re: In a > > letter to the US White House, ACM opposes free distribution of > > peer-reviewed journal articles) > > Message-ID: <20191224172138.GC32219 at traveler> > > Content-Type: text/plain; charset=us-ascii > > > > Dear all, > > I'd like to contribute to this very interesting thread by offering > > some > > background on the Open Acces debate, that started almost a quarter of a > > century > > ago, and that may help younger collegues to get an idea of why we ended up > > where we are today. > > > > It's important to recall that, no matter the official statements one can > > read > > here and there, the goals of researchers on one side, and traditional > > scientific > > publishers on the other, *have always been quite different*: researchers > > wanted > > access to publications as broad and efficient as possible, and a means to > > signal > > excellence of their work; publishers wanted to run a healthy business, and > > a > > monopoly on their own areas (via mandatory copyright transfer and > > ownership of > > the publication trademarks). > > > > It just turns out that in the pre-Internet era what publishers offered also > > satisfied the goals of researchers: the difficult divorce started some > > twenty > > years ago, with the generalisation of the Internet, when sharing papers on > > the > > Web became a *more efficient* means of distibution than having copies of > > journals or proceedings sent to thousands of libraries all over the world, > > and > > the publishers' own goals started to become an obstacle to the free > > dissemination of science. > > > > Unfortunately, before this divorce started, northern emisphere researchers > > (*), as > > well as funding agencies, had broadly agreed to the conditions imposed by > > publishers, in particular mandatory transfer to them of exclusive > > copyright, > > which were not necessary to achieve the researcher's goals. > > > > This had quite disastrous consequences. > > > > On the one hand, it created a gigantic legacy problem: even if we all > > started > > publishing Open Access today, what about the hunderds of thousands of > > articles > > published over the past decades, that are still behind a paywall? > > > > On the other hand, and maybe even worse, it accustomed all the players > > (including governments and funding agencies) to accept the idea that the > > publishing houses actually *own* the copyright to the articles we write, > > while > > they actually *extort* this copyright from the authors themselves, by > > forcing > > them to relinquish their rights in order to be published, a provision that > > is in > > direct violation of the spirit of copyright itself (in France, it is > > actually in > > violation of the letter of copyright law, see article L.131-4 of the CPI). > > > > You can find a trace of this line of thinking in the wording used in the > > letter > > that started this whole thread, where it mentions "intellectual property" > > that belongs to the publishers. > > > > A broader and longer analysis of what is at stake can be found in this > > early > > account I wrote 15 years ago, when some of us had the illusion that Open > > Access > > was going to win soon : > > > > Scientific Publications: The Role of Public Administrations in The ICT > > Era, Upgrade, 2006 > > + available at: http://www.dicosmo.org/FSP/FATOS-Upgrade-03-2006.pdf > > + french original version available at > > http://www.dicosmo.org/FSP/FreeAccessToScience.pdf > > > > Cheers > > > > -- > > Roberto > > > > (*) in Latin America the situation is quite different > > > > ------------------------------------------------------------------ > > Computer Science Professor > > (on leave at INRIA from IRIF/University Paris Diderot) > > > > Director > > Software Heritage https://www.softwareheritage.org > > INRIA > > Bureau C328 E-mail : roberto at dicosmo.org > > 2, Rue Simone Iff Web page : http://www.dicosmo.org > > CS 42112 Twitter : http://twitter.com/rdicosmo > > 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 > > ------------------------------------------------------------------ > > > > GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 > > > > > > > > ------------------------------ > > > > Subject: Digest Footer > > > > _______________________________________________ > > Types-list mailing list > > Types-list at LISTS.SEAS.UPENN.EDU > > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > > > > > > ------------------------------ > > > > End of Types-list Digest, Vol 121, Issue 8 > > ****************************************** > > > > > -- > Valeria de Paiva > http://vcvpaiva.github.io/ > http://www.cs.bham.ac.uk/~vdp/ From roberto at dicosmo.org Sat Dec 28 05:23:47 2019 From: roberto at dicosmo.org (Roberto Di Cosmo) Date: Sat, 28 Dec 2019 11:23:47 +0100 Subject: [TYPES] Types-list Digest, Vol 121, Issue 8 In-Reply-To: References: Message-ID: <20191228102347.GB22709@traveler> Thanks a lot Valeria for sharing this! I did not know of this inspiring decision of ACL, and I find it a very interesting data point in the design space for future Open Access publications. Looking at the FAQ of the ACL Anthology, one learns that the Anthology itself is entirely volunteer based, but I could not find information on how the rest of the cost for handling the editorial process is funded: does it just come from the conference sponsor money/registration fees? All the best -- Roberto On Wed, Dec 25, 2019 at 06:58:09AM -0800, Valeria de Paiva wrote: > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > dear all, > I'd like to add my 2 cents to this thread, replying to this bit of > Roberto's message. > > >On the one hand, it created a gigantic legacy problem: even if we all > started > publishing Open Access today, what about the hundreds of thousands of > articles > published over the past decades, that are still behind a paywall? > > II would like to volunteer the information that the ACL the Association for > Computational Linguistics > https://www.aclweb.org/anthology/ > did exactly this. It made all the articles in their main conferences free > and open access; and then they went back and > digitalized and open-sourced their past! they made the already published > papers also open source. > Now if the computational linguists can do it, I fail to see why we cannot. > To me, it seems a lack of will from our learned societies. But I don't know > the numbers, of course. > > Happy holidays! > Valeria > > On Wed, Dec 25, 2019 at 1:28 AM > wrote: > > > Send Types-list mailing list submissions to > > types-list at LISTS.SEAS.UPENN.EDU > > > > To subscribe or unsubscribe via the World Wide Web, visit > > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > > or, via email, send a message with subject or body 'help' to > > types-list-request at LISTS.SEAS.UPENN.EDU > > > > You can reach the person managing the list at > > types-list-owner at LISTS.SEAS.UPENN.EDU > > > > When replying, please edit your Subject line so it is more specific > > than "Re: Contents of Types-list digest..." > > > > > > Today's Topics: > > > > 1. Re: In a letter to the US White House, (Robert Rand) > > 2. Re: In a letter to the US White House, (Gabriel Scherer) > > 3. Open Access: a bit of background (Was: Re: In a letter to > > the US White House, ACM opposes free distribution of > > peer-reviewed journal articles) (Roberto Di Cosmo) > > > > > > ---------------------------------------------------------------------- > > > > Message: 1 > > Date: Sun, 22 Dec 2019 21:11:42 -0500 > > From: Robert Rand > > To: Peter Selinger > > Cc: Types list > > Subject: Re: [TYPES] In a letter to the US White House, > > Message-ID: > > < > > CAGKXo9deS4hC+Nj_anQ9de3nZ4KfH53Mg6bAPpOhhvgKtadHkw at mail.gmail.com> > > Content-Type: text/plain; charset="UTF-8" > > > > I agree with Peter and Gabriel. The ACM seems intent on charging the same > > amount per article and just shifting around who pays for it: from readers > > to authors to conference-goers to (in their newest scheme) universities as > > part of some sort of general tax. And we shouldn't be paying it. > > > > I would add that for publishing conference proceedings, EPTCS (which > > publishes the conference proceedings for Linearity, ICLP, QPL and dozens of > > other conferences) is a great option. EPTCS is free to everyone, and it > > would be nice if more programming languages conferences switched to using > > it. > > > > On Sun, Dec 22, 2019 at 12:38 PM wrote: > > > > > [ The Types Forum, > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > ] > > > > > > I was recently invited to the editorial board of a new ACM journal. I > > > refused to become an editor, on the grounds that the journal is not > > > open access. > > > > > > I feel that as editors, we volunteer our time, and I have decided to > > > do so only for journals that are truly open access. And by "open > > > access", I do not mean some "gold" model based on author charges > > > (currently $1300-$1700 per ACM journal article, see > > > https://www.acm.org/publications/openaccess). I mean journals that are > > > free for both authors and readers. > > > > > > There are already many such journals, and they are usually > > > community-run. Examples include Logical Methods in Computer Science > > > (https://lmcs.episciences.org/), Compositionality > > > (https://www.compositionality-journal.org/), and Quantum > > > (https://quantum-journal.org/). There is relatively little in the way > > > of interested groups of individuals starting new such journals. > > > > > > I think the best way to get rid of unreasonable publishing models is > > > not to participate in them. > > > > > > -- Peter > > > > > > Jonathan Aldrich wrote: > > > > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > > > gabriel.scherer at gmail.com> > > > > wrote: > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > > > ] > > > > > > > > > > Dear types-list and SIGPLAN, > > > > > > > > > > I have long been of the opinion that our scientific publications > > > > > should be Open Access, and that editors should not request more than > > > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > > > per article). In particular, I believe that copyright transfer > > > > > agreements, as imposed by most editors including the ACM, is deeply > > > > > unethical: the publishers are not the authors of our scientific > > > > > production and they should not force us to give our copyright to > > > > > them. A non-exclusive publishing agreement should be enough. > > > > > > > > > > Whether or not you agree with this position, you may be interested in > > > > > the content of the following letter to the US White House that > > > > > a coalition of scientific publishers, *including the ACM*, signed and > > > > > support. > > > > > > > > > > > > > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > > > press release from the coalition of editors: > > > > > > > > > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > > > > > (This letter was written in the context of a proposed US legislation > > > > > to force more scientists to publish their work in Fair Open Access > > > > > venues. I haven't been able to find a precise link to a discussion of > > > > > this proposed legislation.) > > > > > > > > > > The following parts of the letter co-signed by the ACM are > > > > > particularly juicy: > > > > > > > > > > > [We] have learned that the Administration may be preparing to step > > > > > > into the private marketplace and force the immediate free > > > distribution > > > > > > of journal articles financed and published by organizations in the > > > > > > private sector, including many non-profits. This would effectively > > > > > > nationalize the valuable American intellectual property that we > > > > > > produce and force us to give it away to the rest of the world for > > > > > > free. > > > > > > > > > > > This mandate [...] would make it very difficult for most American > > > > > > publishers to invest in publishing these articles. As a > > consequence, > > > > > > it would place increased financial responsibility on the government > > > > > > through diverted federal research grant funds or additional monies > > > > > > to underwrite the important value added by publishing. In the > > coming > > > > > > years, this cost shift would place billions of dollars of new and > > > > > > additional burden on taxpayers. > > > > > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > > > its weird views (for example, that it really does cost $700 or $900 > > to > > > > > publish an article as Open Source) is good for our research > > > > > community. It do not see how this argument is compatible with the ACM > > > > > signing this letter. > > > > > > > > > > I believe that many of our activities, which we collectively trained > > > > > ourselves to see as harmless administrative details of our research > > > > > work, are in fact empowering the ACM to make those claims. Should we > > > > > accept to give away our copyright, or pay? unreasonable > > > > > Gold Access author processing charges (APCs)? > > > > > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > > > shielding many of us from paying APCs. But many of the smaller > > > > > conferences, symposiums or workshops in our community whose > > > > > proceedings are handled by the ACM are still limited to "pay > > > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > > > copyright transfer as the only free choice, which is effectively > > > > > keeping those proceedings Closed-Access. > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------ > > > > Message: 2 > > Date: Mon, 23 Dec 2019 16:28:35 +0100 > > From: Gabriel Scherer > > To: Peter Selinger > > Cc: Jonathan Aldrich , Types list > > > > Subject: Re: [TYPES] In a letter to the US White House, > > Message-ID: > > > EOhoj9w1tUWMOHBrFSEfnoT4ri6L+HtmhtymUMww at mail.gmail.com> > > Content-Type: text/plain; charset="UTF-8" > > > > I strongly agree with Peter: the most effective action against unreasonable > > publishing models is to not participate in them. We are very fortunate to > > work in a community where many sensible Open Access venues exist, so that > > this decision comes at little to no personal cost. > > > > I created a webpage to document this policy (tentatively named "Keeping my > > work Open"). To evaluate the impact on my work since I started applying it > > in 2017, I listed venues to which I did or did not participate due to the > > policy. > > > > http://gasche.info/open_access.html > > > > The summary is that this principle is not limiting: most venues in > > Programming Languages research have sensible policies (I include PACMPL, > > thanks to the generous support of SIGPLAN to fund APCs, and ETAPS since > > 2018), and that the Closed Access venues I encountered had fairly direct > > replacements in all cases. > > > > On Sun, Dec 22, 2019 at 7:37 AM wrote: > > > > > I was recently invited to the editorial board of a new ACM journal. I > > > refused to become an editor, on the grounds that the journal is not > > > open access. > > > > > > I feel that as editors, we volunteer our time, and I have decided to > > > do so only for journals that are truly open access. And by "open > > > access", I do not mean some "gold" model based on author charges > > > (currently $1300-$1700 per ACM journal article, see > > > https://www.acm.org/publications/openaccess). I mean journals that are > > > free for both authors and readers. > > > > > > There are already many such journals, and they are usually > > > community-run. Examples include Logical Methods in Computer Science > > > (https://lmcs.episciences.org/), Compositionality > > > (https://www.compositionality-journal.org/), and Quantum > > > (https://quantum-journal.org/). There is relatively little in the way > > > of interested groups of individuals starting new such journals. > > > > > > I think the best way to get rid of unreasonable publishing models is > > > not to participate in them. > > > > > > -- Peter > > > > > > Jonathan Aldrich wrote: > > > > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > > > > > > > I agree with Gabriel. Furthermore, I think we should do something. > > > > > > > > On Fri, Dec 20, 2019 at 8:54 AM Gabriel Scherer < > > > gabriel.scherer at gmail.com> > > > > wrote: > > > > > > > > > [ The Types Forum, > > > http://lists.seas.upenn.edu/mailman/listinfo/types-list > > > > > ] > > > > > > > > > > Dear types-list and SIGPLAN, > > > > > > > > > > I have long been of the opinion that our scientific publications > > > > > should be Open Access, and that editors should not request more than > > > > > a fair price (cost of publication, which Dasgtuhl estimates at $60 > > > > > per article). In particular, I believe that copyright transfer > > > > > agreements, as imposed by most editors including the ACM, is deeply > > > > > unethical: the publishers are not the authors of our scientific > > > > > production and they should not force us to give our copyright to > > > > > them. A non-exclusive publishing agreement should be enough. > > > > > > > > > > Whether or not you agree with this position, you may be interested in > > > > > the content of the following letter to the US White House that > > > > > a coalition of scientific publishers, *including the ACM*, signed and > > > > > support. > > > > > > > > > > > > > > > > > > > > > > > > > https://presspage-production-content.s3.amazonaws.com/uploads/1508/coalitionletteropposinglowerembargoes12.18.2019-581369.pdf > > > > > press release from the coalition of editors: > > > > > > > > > > > > > > > https://newsroom.publishers.org/researchers-and-publishers-oppose-immediate-free-distribution-of-peer-reviewed-journal-articles > > > > > > > > > > (This letter was written in the context of a proposed US legislation > > > > > to force more scientists to publish their work in Fair Open Access > > > > > venues. I haven't been able to find a precise link to a discussion of > > > > > this proposed legislation.) > > > > > > > > > > The following parts of the letter co-signed by the ACM are > > > > > particularly juicy: > > > > > > > > > > > [We] have learned that the Administration may be preparing to step > > > > > > into the private marketplace and force the immediate free > > > distribution > > > > > > of journal articles financed and published by organizations in the > > > > > > private sector, including many non-profits. This would effectively > > > > > > nationalize the valuable American intellectual property that we > > > > > > produce and force us to give it away to the rest of the world for > > > > > > free. > > > > > > > > > > > This mandate [...] would make it very difficult for most American > > > > > > publishers to invest in publishing these articles. As a > > consequence, > > > > > > it would place increased financial responsibility on the government > > > > > > through diverted federal research grant funds or additional monies > > > > > > to underwrite the important value added by publishing. In the > > coming > > > > > > years, this cost shift would place billions of dollars of new and > > > > > > additional burden on taxpayers. > > > > > > > > > > In my discussion with many of us, I regularly hear that the ACM is > > > > > "not evil" (the SIGPLAN, of course, is pure good!) and that placating > > > > > its weird views (for example, that it really does cost $700 or $900 > > to > > > > > publish an article as Open Source) is good for our research > > > > > community. It do not see how this argument is compatible with the ACM > > > > > signing this letter. > > > > > > > > > > I believe that many of our activities, which we collectively trained > > > > > ourselves to see as harmless administrative details of our research > > > > > work, are in fact empowering the ACM to make those claims. Should we > > > > > accept to give away our copyright, or pay? unreasonable > > > > > Gold Access author processing charges (APCs)? > > > > > > > > > > ?: The SIGPLAN decision to cover APC costs for PACMPL articles is > > > > > shielding many of us from paying APCs. But many of the smaller > > > > > conferences, symposiums or workshops in our community whose > > > > > proceedings are handled by the ACM are still limited to "pay > > > > > $900" (or "pay $25 per page") as the only Open Access option, with > > > > > copyright transfer as the only free choice, which is effectively > > > > > keeping those proceedings Closed-Access. > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------ > > > > Message: 3 > > Date: Tue, 24 Dec 2019 18:21:38 +0100 > > From: Roberto Di Cosmo > > To: Gabriel Scherer , Types list > > > > Subject: [TYPES] Open Access: a bit of background (Was: Re: In a > > letter to the US White House, ACM opposes free distribution of > > peer-reviewed journal articles) > > Message-ID: <20191224172138.GC32219 at traveler> > > Content-Type: text/plain; charset=us-ascii > > > > Dear all, > > I'd like to contribute to this very interesting thread by offering > > some > > background on the Open Acces debate, that started almost a quarter of a > > century > > ago, and that may help younger collegues to get an idea of why we ended up > > where we are today. > > > > It's important to recall that, no matter the official statements one can > > read > > here and there, the goals of researchers on one side, and traditional > > scientific > > publishers on the other, *have always been quite different*: researchers > > wanted > > access to publications as broad and efficient as possible, and a means to > > signal > > excellence of their work; publishers wanted to run a healthy business, and > > a > > monopoly on their own areas (via mandatory copyright transfer and > > ownership of > > the publication trademarks). > > > > It just turns out that in the pre-Internet era what publishers offered also > > satisfied the goals of researchers: the difficult divorce started some > > twenty > > years ago, with the generalisation of the Internet, when sharing papers on > > the > > Web became a *more efficient* means of distibution than having copies of > > journals or proceedings sent to thousands of libraries all over the world, > > and > > the publishers' own goals started to become an obstacle to the free > > dissemination of science. > > > > Unfortunately, before this divorce started, northern emisphere researchers > > (*), as > > well as funding agencies, had broadly agreed to the conditions imposed by > > publishers, in particular mandatory transfer to them of exclusive > > copyright, > > which were not necessary to achieve the researcher's goals. > > > > This had quite disastrous consequences. > > > > On the one hand, it created a gigantic legacy problem: even if we all > > started > > publishing Open Access today, what about the hunderds of thousands of > > articles > > published over the past decades, that are still behind a paywall? > > > > On the other hand, and maybe even worse, it accustomed all the players > > (including governments and funding agencies) to accept the idea that the > > publishing houses actually *own* the copyright to the articles we write, > > while > > they actually *extort* this copyright from the authors themselves, by > > forcing > > them to relinquish their rights in order to be published, a provision that > > is in > > direct violation of the spirit of copyright itself (in France, it is > > actually in > > violation of the letter of copyright law, see article L.131-4 of the CPI). > > > > You can find a trace of this line of thinking in the wording used in the > > letter > > that started this whole thread, where it mentions "intellectual property" > > that belongs to the publishers. > > > > A broader and longer analysis of what is at stake can be found in this > > early > > account I wrote 15 years ago, when some of us had the illusion that Open > > Access > > was going to win soon : > > > > Scientific Publications: The Role of Public Administrations in The ICT > > Era, Upgrade, 2006 > > + available at: http://www.dicosmo.org/FSP/FATOS-Upgrade-03-2006.pdf > > + french original version available at > > http://www.dicosmo.org/FSP/FreeAccessToScience.pdf > > > > Cheers > > > > -- > > Roberto > > > > (*) in Latin America the situation is quite different > > > > ------------------------------------------------------------------ > > Computer Science Professor > > (on leave at INRIA from IRIF/University Paris Diderot) > > > > Director > > Software Heritage https://www.softwareheritage.org > > INRIA > > Bureau C328 E-mail : roberto at dicosmo.org > > 2, Rue Simone Iff Web page : http://www.dicosmo.org > > CS 42112 Twitter : http://twitter.com/rdicosmo > > 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 > > ------------------------------------------------------------------ > > > > GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3 > > > > > > > > ------------------------------ > > > > Subject: Digest Footer > > > > _______________________________________________ > > Types-list mailing list > > Types-list at LISTS.SEAS.UPENN.EDU > > https://LISTS.SEAS.UPENN.EDU/mailman/listinfo/types-list > > > > > > ------------------------------ > > > > End of Types-list Digest, Vol 121, Issue 8 > > ****************************************** > > > > > -- > Valeria de Paiva > http://vcvpaiva.github.io/ > http://www.cs.bham.ac.uk/~vdp/ -- Roberto Di Cosmo ------------------------------------------------------------------ Computer Science Professor (on leave at INRIA from IRIF/University Paris Diderot) Director Software Heritage https://www.softwareheritage.org INRIA Bureau C328 E-mail : roberto at dicosmo.org 2, Rue Simone Iff Web page : http://www.dicosmo.org CS 42112 Twitter : http://twitter.com/rdicosmo 75589 Paris Cedex 12 Tel : +33 1 80 49 44 42 ------------------------------------------------------------------ GPG fingerprint 2931 20CE 3A5A 5390 98EC 8BFC FCCA C3BE 39CB 12D3