[sigcomm] industrial participation ... Re: another approach

[jeffmogul@acm.org]

Since we've shift topics a little, I'll jump in.  I've tilted
twice at this particular windmill (i.e., trying to get
non-academic people to publish research-like papers).  Once was
as the chair of the second and last "Workshop on Industrial
Experiences with Systems Software"
<http://www.usenix.org/events/wiess02/>, and once as co-chair of
the SIGCOMM 2003 NICELI workshop
<http://www.acm.org/sigs/sigcomm/sigcomm2003/workshop/niceli/>.

Neither event attracted a lot of submissions, although I
think in the end they were both valuable for the attendees.  It
definitely takes a serious commitment on the part of the PC chair
and members to aggressively solicit papers from industrial authors
(excluding the major corporate research labs).  It's not like
SIGCOMM or HotNets or IMC, which always seem to have large pools of
submissions.

In the case of WIESS, I think we decided that the well was
basically dry, partly because the scope was too narrow.  NICELI
also had a rather narrow scope, but because it was a hot topic at
the time (RDMA, basically) the result was more coherent.  Also, I
think NICELI succeeded in part because it was co-located with
SIGCOMM, which meant that both attendees and submitters had other
incentives to participate.  (WIESS was co-located with OSDI,
which gave us a good audience but somehow didn't result in
submissions.)

My recommendation would be to consider co-locating a workshop on
industrial networking innovation, engineering, and operation at the
annual SIGCOMM conference.  I'd discourage pure-research
submissions (but academic-practitioner collaborations would be
great).  I suspect there are enough potential authors (if the
scope is broad enough) outside the usual academic + research-lab
world to make this viable, although it will take some effort to
solicit enough papers.  It's crucial that the organizers not try
to impose the same acceptance criteria as one would expect for a
true research conference/workshop, or else we would just end up
with Yet Another one of those.

-Jeff