[TYPES/announce] Two phase reviewing for POPL; a response

[James Cheney]

Hi,
Quoting from the proposal:

| * Other conferences are moving to a year-round refereeing process
| closer to that used by journals; for instance VLDB is now linked to
| a journal PVLDB.  The two-phase proposal yields similar benefits,

| while ensuring focus and a bound on effort.


Though I don't have direct experience with the "journal track" of PVLDB, I
have heard positive experiences from others and know some of the people who
set it up.  I think the above description is inaccurate and it's unclear
that the POPL SC proposal has the claimed advantages.  So I'd like to
describe how the VLDB/PVLDB system currently works.

At present, VLDB retains an ordinary conference organization.  All papers
accepted to VLDB are published in "Proceedings of VLDB", a journal, and all
authors of such papers are given a slot to present.  In addition, there is a
"journal track": a paper can be submitted anytime during the year.  It is
supposed to be reviewed within a set period (similar to the time for
conference review) with a hard decision after the second round - very
similar to the 2-round proposal for POPL, but not anchored to the conference
deadline.

All accepted papers, whether submitted via the traditional conference review
process or journal track, can be presented at the next suitable instance of
the conference.  (I believe for journal track papers this is optional).  The
PVLDB folks anticipate that other conferences might participate in the
system, so eventually there might be a choice of venues for accepted
journal-track papers.  Also, all papers have the same length and are cited
as journal papers in PVLDB.

I believe both proposals have the properties that are claimed as advantages
only for the POPL proposal:
1.  the ordinary conference submission process is preserved, providing
focused deadlines for the majority of papers (and authors who prefer
deadlines and predictability)
2.  the additional journal track provides reasonable bounds on PC member and
external reviewer effort

Indeed, the PVLDB process may be less effort for PC members, since the
reviewing process for the conference track is unchanged.  The additional
effort is needed from journal track reviewers and editors, who are not
necessarily PC members.

On the other hand, the VLDB system was a response to a different problem:
VLDB and other major DB conferences have already grown large (3-4 parallel
research tracks, PCs with > 100 members) and the reviewing process is
perceived as random; DB systems researchers have co-evolved with this system
to submit many papers every year with lots of roll-over (of both good and
bad papers).  So I'm not claiming the VLDB approach is a solution to the
problem identified here.

I do think it avoids many of the problems Simon perceives with the proposed
change, though.  It also has the advantage that it's relatively incremental
- instead of moving everyone's deadlines and giving the whole PC
(potentially) more work, it is up to the author to decide whether to submit
early via the journal track or through the traditional, more predictable
conference process.

--James
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.seas.upenn.edu/pipermail/types-announce/attachments/20100112/035d9233/attachment.htm