Monday, July 07, 2008

COMSOC notification

Program co-chair Ulle Endriss and I finally finish choosing papers for presentation at COMSOC-2008. Thanks to the program committee members for their work, and also to everyone who submitted papers. We accepted 36 out of 55 submissions (not including a handful of off-topic or duplicate submissions that were deleted at the outset). Authors have now been notified of the decisions on their papers. Here are some thoughts about picking papers for this particular meeting.

The process of choosing papers was done in computer-science style, i.e. with the aid of a program committee of established researchers in the field, who reviewed the papers in detail. We obtained three reviews per paper, which were mostly passed back to the authors. This approach is of course geared towards a system whereby you are looking mainly for technical merit of papers.

Computer science conferences are treated as publication forums, but in economics (and I believe in most other disciplines) conferences are much more informal. There is a proceedings but papers may be in a preliminary state. Fundamentally, COMSOC is a meeting, not a publication forum, and we took papers that appeared elsewhere. But, we compromised a bit with the Computer Science way of doing it.

In CS conferences, at least the ones where I have reviewed papers, the aim is to select (from the papers deemed relevant to the meeting) the ones of highest technical quality. The discussion of which papers to select is aimed at a rank-ordering of the submissions, in terms of quality, and the "best" ones are taken. For COMSOC, we had that bias, but we also aimed for representation of different sub-topics, and had a bias towards younger authors and paper that had not so far been presented anywhere else (actually, for CS conferences, that last aspect is not so much a bias as a requirement; since they are publication forums a paper is disqualified if it was given at a previous conference).

Grounds for rejection included "not being about computation", which applied to a sizeable minority of papers, including some quite interesting ones. (You have to be caseful to make a broad interpretation of what it means to be about computation, however.) "not being about social choice" was only applied to one paper that I know of, where it seemed to be more about a standard optimization problem that was being discussed in the context of a particular social choice setting.

I liked the approach of attempting to build a program that we believe should develop new interactions and attract interesting participants, rather than just select for technical quality. There's something quite liberating about being able to admit that it's OK to take one paper and reject a "better" one.

Due to the compromise that we adopted, there is a risk that some contributors from CS, as well as come from (say) economics, especially ones whose papers were rejected, will be taken aback by the approach to selecting papers. An obvious topic for the business meeting is whether we want a subsequent COMSOC to veer towards one or the other type of conference.

No comments: