[sigcomm] attendance policies for SIGCOMM-affiliated events

[Jennifer Rexford]

Joe,

> The key question is _why_ random isn't good enough, and why limit
> attendance at all?

I'll take a stab at this, despite my personal preference for open 
attendance.  Why limit attendance?  Because attendance, along with other 
factors such as venue selection, have a significant impact on the amount 
of discussion that ensues in a workshop.  I attended the MIT HotNets 
(because I was presenting a paper) and it was, in my opinion, the best 
workshop or conference I ever attended.  The room was small, and 
shallow, and the mingling area outside the room was small.  We didn't 
need microphones.  People interupted during the presentations, and talks 
engendering discussion ran long.  Attendance was by no means the only 
factor here, but it was undoubtedly a factor.  Would I want other people 
in the community to have the benefit of this experience?  You bet.

Yet, I'm not able to say what the maximum number is where this dynamic 
starts to break down, and I appreciate that the breaking point does not 
depend only on size.  For example, a larger room, or a room with poor 
acoustics, or a room that opens into a distracting place where people 
want to mingle instead of attending the talks, etc. all detract from the 
experience.  Fortunately, all of these factors can be controlled in 
venue selection, and the organizers take on the responsibility for 
handling that, so these issues are orthogonal to our discussion here. 
Attendance size, and the makeup of the attendees, are the only remaining 
issues before us.

So, what are the trade-offs between random and other methods?  Random is 
clearly fair, which is an important goal for a professional society. 
Would random attendance generate the same level of discussion as other 
methods?  I don't know, but my inclination is that it might not.  For 
example, if you did "random amongst authors of submitted papers" or 
"random among authors of papers the PC viewed as above some bar -- i.e., 
excluding papers that were clearly written just to be eligible to 
attend", you might get folks more likely to contribute actively to the 
discourse.  Of course, such methods (like any others) are just 
heuristics, as you can't easily control whether people will be active, 
thoughtful participants at a workshop, but these kinds of methods might 
be reasonably viewed as more likely to engender technical discussion.

Then, sprinkle in the fact that we really want to have students 
participate in all of our events, because they are the lifeblood of our 
community, and they (like the rest of us) learn on the job.  So, I could 
see forgoing "random" in favor of a policy that tried to get a mix of 
students (e.g., as authors of strong submissions that weren't accepted).

Anyway, where am I heading with this?  I think we have a classic case of 
competing priorities here.  Personally, I think the community is 
interested in an open event where hot topics are presented and 
discussed, and I lean toward openness, even if a larger event inevitably 
decreases the amount of discourse somewhat.  Yet, I can see the points 
of the other side on this, in terms of wanting to limit attendance to 
keep the level of discourse high.  I (again, personally, as opposed to 
trying to dictate policy) can see the SIG sponsoring a few limited 
attendance events, as long as the policy is transparent and fair -- 
where I realize that fair is a slippery word.

-- Jen <-- with SIG chair hat off