[TYPES] Diamond open access, cost and sustainability model: a JOSS example

[Gabriel Scherer]

Dear list,

I was pointed off-list to the following interesting article from 2020, the
first attempt by ACM to provide some sort of public financial information.

  CACM: ACM Publications Finances

https://cacm.acm.org/magazines/2020/5/244322-acm-publications-finances/fulltext

The report is interesting, and it suggests that the costs of publishing at
ACM are quite high. Some details below. (Those are all 2019 numbers)

- There is a staggering difference in per-article cost between "ACM
journals" and "conference proceedings". According to this data, ACM Journal
cost on average $1500 per article, while conference proceedings cost of
$151 per article. In particular, the data reports that each ACM journal
article spends $333 (on average) on "composition and copy-editing".
  My uninformed guess would be that PACML, while being nominally a journal,
is still handled by conference-proceedings processes: I have published in
conference proceedings and then PACMPL, and not observed any change that
would correspond to a ten-fold cost increase.

- ACM reports a conference proceeding cost of $151 per article. In the blog
post, they report it as $410, but in fact much of that figure corresponds
to ACM revenue that they give to SIGs and include in the "cost of
publication" of the SIG. (It's great that ACM makes revenue and that they
fund the SIGs, but we are trying to understand publishing costs.)

- None of the figure above include the ACM Digital Library (web hosting +
long-term archiving), whose cost in 2019 were massive: $299 per conference
proceeding publication on average. The post has more details on that: 2019
was right after they launched the "new DL" platform, so a share of these
costs are fixed and will not recur on following years. But they also expect
to host more video content (indeed), so some other costs will increase.

  Note that arxiv has costs of $14 per article, and Zenodo offers long-term
archiving of gigabyte-large content for free as a public service (
https://help.zenodo.org/ ; this is supported as a "drop in the bucket" of
CERN physicists costs archiving petabytes of experiment data. We use Zenodo
to host the artifacts submitted to the Artifact Evaluation processes of
several SIGPLAN conferences. )
  ACM should let authors choose to publish on the ACM DL *or* on
Arxiv+Zenodo; the people who see value in ACM DL could choose this option,
and the others would vastly reduce hosting/archiving costs (from $299 to
$14).

- The revenue/costs numbers for ACM ICPS are interesting (International
Conference Proceeding Series: as a non-ACM conference you can contract ACM
to publish your proceedings, for a small per-paper price, in exchange for
forcing your authors to give up their copyright to the ACM; several
conferences of our community do this, for example PPDP). In 2019 they
brought $362K of revenue in publication fees, for about $250K of publishing
costs¹.
  And the ICPS publishing fees are pretty reasonable! See the fee structure
at https://www.acm.org/publications/icps-series : for one edition of
proceedings, you pay $750 of fixed costs for up to 30 articles, plus $20
for each paper above the 30th.
  If ACM makes a net profit with just those fees (the revenue figure is
just for publication fees, it does not include subscriptions, pay-per-view
etc.), this means that our ACM conferences could pay *exactly this price*
for (Open Access) proceeding publications and not cost ACM any money, in
fact bring revenue. (This only covers publishing costs, not ACM DL costs.)

¹:  ACM gives a figure of $215K ofr ICPS publishing costs, but it has not
estimated overhead expense, so the figure is an under-estimate. We can
estimate this cost. ICPS publishes on average 255 papers per year since
2002; optimistically assuming 1000 papers in 2019, if the overhead costs
are proportional to the other conference proceeding overehead cost, this
adds another $34K, for a total cost of $250K.

On Sun, May 30, 2021 at 6:52 PM Gabriel Scherer <gabriel.scherer at gmail.com>
wrote:

> Dear list,
>
> Today I found out about JOSS, the Journal of Open Source Software (
> https://joss.theoj.org/ ), an interesting journal in itself, which has a
> stunning "Cost and sustainability model" webpage section:
>   https://joss.theoj.org/about#costs
>
> For more stunning details, go read their more detailed blog post, "Cost
> models for running an online open journal" : )
>
> http://blog.joss.theoj.org/2019/06/cost-models-for-running-an-online-open-journal
>
> (Meanwhile in ACM land, we are still waiting for basic financial
> transparency on paper publishing costs -- not that, say, ETAPS or JFP are
> doing any better.
> LIPIcs describes how they calculated their publishing costs at
> https://www.dagstuhl.de/en/publications/lipics/processing-charge/ , and
> LMCS ( https://lmcs.episciences.org/ ) is now using a publicly-funded OA
> publishing platform, so they may actually have no costs at all.)
>
> Cheers
>