Eva Tardos

To summarize in a single sentence, I think this is a really bad idea, and would like to strongly urge you NOT to do this. If you think there is a need for such a venue, I see nothing wrong in suggesting to have one. But why kill a really good meeting, like STOC.
For a longer version: you are comparing to how scientific meetings work in other science fields, and basically suggest that we adapt to how meetings work at other science fields. Each field has its own culture, and I am sure the way the physics community works is well supported by the way they run their meetings. However, we are a subfield of computer science (as an ACM SIG), and it is best for us if our meetings are set up similar to meetings in other CS areas. This certainly helps those of us in academic computer science departments, where the people from the other subareas of computer science are our immediate colleagues. Typically CS subfields are organized around selective conferences like FOCS and STOC. The systems people has SOSP and OSDI, the database people have SIGMOD and VLDB, vision people have CVPR and ICCV, the machine learning area has ICML and NIPS, programming languages has POPL and PLDI, graphics has SIGGRAPH, etc.
You correctly point out that other scientific communities are not organized around such meetings. The math community isn’t, physics mostly isn’t, chemistry isn’t, and other engineering fields aren’t either. But CS fields are. There are huge benefits of having a community and meeting structure that is more similar to other subfield of computer science, than to math. I do admit that scientific computing is more like math and not CS. But they increasingly do get questioned whether they should belong to math or to CS. The majority of academics in our community are in CS departments (as are all but one members of this EC). It serves our field incredibly well that we are well integrated in the broader computer science community. By being in CS departments (or CS groups at companies) we get involved and impact issues that the broader community thinks about. This is important to each of us, and certainly important to our field. I think the proposed more math-like arrangement of the new STOC will hurt our ability to do this. I can still remember the not-so distant past when many theoreticians felt that their area was questioned in their own department. Do you remember the Karp committee report on the future of theoretical computer science from 1997. We are at a much better point today. Changing our meeting structure closer to the math community is a pretty big step in the wrong direction.
I assume you will argue that your proposal kept one such conference FOCS, and that should be enough. I argue that its not enough. As you saw on the list most areas have two and they often have a spring and a fall one (just like we do). The graphics community has one only: SIGGRAPH, and this single highest quality publication venue makes life really hard for graphics people. They focus a lot of their afford all year on the December SIGGRAPH deadline. Waiting 9+ months to get something published at FOCS (or submitted to FOCS) does not seem healthy for the community and does not seem good for science. Systems till relatively recently had only one such opportunity SOSP and OSDI alternating both in the fall. They added the related NSDI in the springs 7 years ago.
Answering your concrete questions, I do feel STOC adequately serves the community. I am sure not perfect, nothing is perfect. Improvement is always possible. But the proposal is a big step in the wrong direction.

No comments:

Post a Comment