Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cwjcc!gatech!ulysses!smb
From: smb@ulysses.homer.nj.att.com (Steven M. Bellovin)
Newsgroups: comp.org.usenix
Subject: Re: suggestions for future conferences
Message-ID: <11210@ulysses.homer.nj.att.com>
Date: 12 Feb 89 20:22:37 GMT
References: <8902090223.AA01916@decwrl.dec.com>
Organization: AT&T Bell Laboratories, Murray Hill
Lines: 47

As other responders have said, the content of the conference is largely
determined by the papers submitted.  For example, for Baltimore Usenix
about 10% of the papers submitted appear (from their titles) to be
related to security.  That more or less guarantees some presentations
on security topics, unless they're all turkeys.  On the other hand, I
don't see any papers on neural networks (though they were explicitly
solicited in the Call for Papers), nor anything on troff and its
friends.

I was on the program committee for the Salt Lake City Usenix (June '84);
let me describe what we did.  We sorted through the papers, and
performed a rough triage:  great paper, might be usable, and obvious
turkey.  We then made another pass on the middle pile, sorting it
further.  Then we looked at what themes we had, and started assigning
papers to different sessions.  We did have two tracks, one in a large
auditorium, and one in a smaller ballroom, so we had to guess which
would be popular vs. which would require interactions.  Panel
discussions, for example, were in the ballroom, so the audience could
heckle (I mean comment) better.  Finally, we filled in the holes with 1
or 2 papers that we felt were good enough, and complemented the other
papers in the session.  We did not accept papers that didn't meet our
standards, though we were hindered in our judgements because
submissions at the time were of abstracts only, not complete papers.

We also got a lot of flack for the track assignments, from authors who
felt they'd been slighted, or from attendees who thought that two
sessions clashed.

Panel discussions are a tricky matter to organize because you want
topics (and speakers) who will disagree, disagree loudly enough to make
it interesting, but politely enough that the conference doesn't start
to resemble alt.flame.  Some people do not like panel discussions
because they leave no permanent record; there's nothing that can be
cited in a later paper.  But they're a good way to present current
opinions.

The committee is currently reviewing the submissions for the Baltimore
Usenix; we'll make our decisions in early March.  If you have any
concrete suggestions (i.e., topics you'd like to see panel discussions
on), please mail them to the committee before March 1:

	  usenet: {ucbvax,decvax,decwrl,seismo}!sun!balt-usenix
	  internet: balt-usenix@sun.com

If you can suggest moderators or panel members, so much the better.

		--Steve Bellovin