Tuesday, August 11, 2009

Rump Sessions at Conferences

The Complexity Conference has almost always had a Rump Session which is where people sign up to give a 10 minute talk on what they are working on. (They didn't have one in 2009 to make room for more talks since there were more submissions.) It can be at any level of development--- open problems, partial results, finished results.

Some conferences usually have rump sessions and some usually do not. By asking around (not always reliable) I have found that STOC, FOCS, SoCG usually do not have rump sessions but that CCC, Crypto, Eurocrypt, AsiaCrypt, AfricaCrypt, TCC (Theory of Cryptography Conference), FSE (Fast Software Encryption) usually do have rump sessions. (A different question, dealt with in Jon Katz's Aug 7 blog is are there too many crypto conferences. That post and the comments on it also list more crypto conferences.) I would like commenters to correct and add to this list.
  1. I do not know where the term Rump Session comes from. I first heard it from Alan Selman at the first CCC business meeting, though he used it as though it was a common term. If you Google the term you get around 19,000 hits, mostly things like 2007 Crypto Rump Session
  2. At a big conference a Rump Session might be harder to organize. That may explain why STOC, FOCS don't have them, but CRYPTO is pretty big and does, while SoCG is small and doesn't.
  3. Early on Rump sessions were on a blackboard. Now many of the Rump sessions are prepared polished presentations. I am tempted to say that this means they are planned ahead of time and can't be things worked on at the conference, but actually people can now whip up polished presentations pretty fast.
  4. Rump sessions are a good way to communicate informal unpolished ideas and ask for help on formalizing or polishing them. And for help on the hard math as well. I think that conferences should have them, though I realize that there may be logisiticaly problems.
  5. Should you share your open problems with others so openly? If you want them solved and don't care if you are the one who solves them then YES. Frank Stephan and Martin Kummer solved an open problem that I proposed at a COLT Rump Session. They got a paper out at the next COLT where I was acknowledged. I was happy that my problem got solved.
  6. Why do some conferences have it and others do not? My guess is the usual one Because we've always had (not had) them. Also, I suspect that the XXX-CRYPTO conferences have them because the first one, CRYPTO, had it.
  7. What do non-theory Comp Sci Conferences do? How about outside of theory? In fields where they do not have prestige conferences perhaps many talks are what we call rump sessions.


  1. OOPSLA added a Lightning Talk session a few years ago. Speakers could talk about anything: their current work, an open problem, a suggestion for the community, a beef with the software world, ... This sort of session is a great way to make unexpected connections.

  2. What is with you and AfricaCrypt? It's like you purposefully picked the most obscure crypto conference out there...

  3. Anon 2: I generated my list of Crypto conferences a while back - I might have just typed CRYPTO into Google and got the list.
    Also, that one I found out DOES have a Rump Session so it fit my point well.

    No bigger point was intended.

    bil g.

  4. Merriam-Webster defines rump as "a small or inferior remnant or offshoot; especially : a group (as a parliament) carrying on in the name of the original body after the departure or expulsion of a large number of its members"

  5. In the book "Makers of the Media Mind" (1990) they write "Several scholars got together at the 1955 AEJ meeting at Boulder, Colorado, in a post-convention meeting that became the first "rump session."

    The book "Human Communication as a Field of Study" (1989) makes the same claim (usuing different words and details).

  6. In my (short) experience attending crypto conferences, the reality seems to be that these rump sessions are only used to advertise finished, submitted results, and not to stimulate conversation or solicit help. This isn't necessarily all bad, but it makes me less inclined to attend: I know I'll see a more complete version eventually anyway, and the conferences are so full as it is that it's an easy session to skip.

    An interesting note: in the TCC rump session, speakers are strongly encouraged to be funny.

  7. In graphics I haven't seen any rump sessions, either at SIGGRAPH or at the smaller conferences (Symposium on Geometry Processing or High Performance Graphics). HPG had keynotes and a panel from people in industry, plus a "where will high performance graphics be in 7 years?" panel, which were helpful for getting a perspective on open problems and trends. SGP didn't really have anything along these lines.