[sigcomm] considerations for reviewing extended papers

[Vern Paxson]

At the last business meeting we discussed the treatment of papers submitted
to SIGCOMM conferences for which shorter papers describing the work in an
earlier phase were previously published.  Should the extended, full-length
papers be evaluated only on their contribution beyond the earlier, short
paper?  If so, is this done taking into account in some fashion that it's
the same set of authors?  Or should the full-length submission be evaluated
as though the previous paper was simply a technical report?

This issue arises because recently SIGCOMM has fostered a number of
venues (the workshops affiliated with the annual conference; HotNets;
extended-abstract submissions to IMC) that solicit short papers.  If
publishing these short papers makes the longer version of the paper
harder to publish in a high-profile forum such as the SIGCOMM annual
conference, then authors will be hesitant to submit, undermining the very
purpose of these venues for trying to open up the SIGCOMM community
to work in earlier stages.

At the meeting, an additional view was expressed (by Emin Gun Sirer,
I believe) that discouraging such preliminary short papers is particularly
problematic for those doing systems work, for which publishing an early
short paper describing the vision of the system can be a highly valuable
means to get early feedback on what will be a major ensuing undertaking.

This is clearly a cultural issue, in that different research communities
apply quite different rules.  In addition, clearly SIGCOMM cannot dictate
how *other* venues (e.g., INFOCOM) will treat longer versions of papers
that appeared earlier in short form in SIGCOMM venues.  However, our goal
is to at least formulate a public statement about SIGCOMM's stance, which
perhaps/hopefully the PCs of such other venues will take into consideration
in some fashion.

At the business meeting I framed a proposal from the Executive Committee
that SIGCOMM adopt a policy of treating earlier, short papers as the same
as technical reports.  However, this issue proved highly contentious
in the ensuing discussion.  One of the themes discussed was that overall
the problem arises because we *still* lack good venues for work in earlier
stages; if our workshops didn't have proceedings, or if they did but
thematically they catered to truly wild/rough work for which the delta to
a full paper is quite large, then the problem would essentially go away.

One of the proposals was for more emphasis on poster presentations.
Currently, these are limited to students at the annual conference.  What
about opening this tract up for broad participation?

This year's SIGCOMM PC chairs used a policy that I would summarize as
"*if* there's a meaningful delta in a SIGCOMM submission beyond the
previous short paper, *then* the submission should be treated as though
the short paper had not appeared".

To spur discussion with an aim towards converging on a policy in this
regard, Jen & I would like to propose that SIGCOMM goes forward with the
policy framed by the Executive Committee: treat earlier, short papers the
same as technical reports.

Do we have a majority of agreement with this?  Or proposals for other
policies to adopt instead?

		Vern