The accepted papers list for FOCS 2014 is now posted online.
I am always amazed by the depth and breadth of works in the TCS community, and this FOCS is no exception. Whether you are a physicist interested in the possibility of general “area law” governing entanglement between different parts of systems, a geometer interested in Gromov’s topological notion of expanders, an optimization expert interested in the latest and greatest in interior point methods, a game theorist interested in Karlin’s longstanding conjecture on convergence of fictitious play, a complexity theorist interested in the latest efforts to separate the determinant from the permanent, or simply a dog owner or triangle lover, you will find something of interest in the conference. And of course FOCS is not just a collection of unrelated papers. A quantum computing expert would want to check the paper on topological expanders, as similar concepts have arose in the context of topological approaches to quantum error correction. An optimization expert might want to understand the convergence of “fictitious play” which is a very natural algorithm for solving linear programs, and of course since STOC 2014 we all know that circuit lower bounds are tightly connected to improving the exponents of algorithms for combinatorial problems. This is just a taste and I could have chosen many other such examples, all strengthening Avi Wigderson’s point why we should all go to talks in areas other than our own.
I was also amazed by the effort reviewers and program committee members have put in the selection process. Conference reviewing sometimes get a bad reputation as being superficial. I did not find this to be the case at all. People have invested an amazing amount of work reading the papers, checking proofs, chasing down references, verifying technical points with the authors and other experts, and generally doing the best job they can to have an informed selection process and assemble the best program we can for the TCS community. I am sure we made mistakes, and the final program, as a product of a committee, is not fully consistent with any particular PC member’s taste, including my own. In particular, there were many submissions that some of us personally found novel and interesting, but were not included in the final program. But I do feel good about the process and believe that while some of our decisions may have been wrong, they were not wrong because we were superficial or lazy or cut corners due to the time pressure. Many times during this process I asked the PC members to go above and beyond what is typically expected, and they have more than risen to this challenge, often making heroic efforts to understand very complex (and sometimes not so greatly written) papers, and trying to get to the bottom of any misunderstanding. I am deeply grateful to them all.
Finally, some statistics. We accepted 70 papers, which is about 26% of the 268-273 submissions (depending on whether you count withdrawn ones). Aside from 9 submissions that were judged to be out of scope and received minimal reviewing, on average each submission had 3.3 reviews and 11.7 comments (including both comments by the PC and short comments/opinions by outside experts that were solicited in addition to the reviews.) Of course these numbers varied greatly based on how much attention and investigation we felt each submission needed and there was also extensive discussion on some of the papers during our two long days of the physical PC meeting. Finally, a very personal statistic for me is that there are about 2800 emails in my “FOCS14” folder. As many past chairs told me, the best thing about this job is that you only do it once…
Windows On Theory 
70 papers is less than the number accepted back in 1990, 1991 or 1992. Why this is considered a desirable situation is beyond me.
And it is more than the number accepted in 2005 and 2007. The acceptance rate is within the range of recent FOCS/STOC conferences (see http://www.lamsade.dauphine.fr/~sikora/ratio/confs.php ), some variability in the number of papers in one direction or the other is to be expected.
So to be clear, you see nothing wrong with the conference being the exact same size as it was 25 years ago, in spite of the rapid growth in the field?
What is the right size for FOCS is a good question, and there are multiple factors to consider. I don’t think it is obvious that a conference has to grow with the field. (In particular, while I didn’t do extensive research, I believe that many top conferences in other areas of CS such as systems and networking also stayed at roughly the same size in the last couple of decades.)
The growth of the field has resulted in many more conferences, which is a good thing but also means people have even more demands on their travel time, and so it seems that we would not be able to add a day or two to FOCS and STOC. And if you grow the conference by adding parallelism, then you are decreasing the chance that people would go to talks in other areas, which is the entire point of this post (and to a large extent the entire point of broad conferences such as FOCS and STOC).
Also, some of this discussion on number of accepted papers seems to assume that we could simply turn a dial and increase the number of accepted papers, but this is actually not so easy. It’s not as if we ranked all papers on some linear scale and then decided to put a threshold somewhere. We discussed in the PC meeting every one of the papers we accepted (and of course also many papers that we didn’t). It is not hard to see that we are already stretching the limits of what you can discuss in a 2 day meeting.
You could drop the physical meeting and PC-wide discussions of all papers but that also causes a qualitative change in the conference, since it would mean that decisions would mostly get made by the specialist reviewers and not by the broader PC.
“I believe that many top conferences in other areas of CS such as systems and networking also stayed at roughly the same size in the last couple of decades.”
This was the case until about five years ago. Since then most of them have moved in the direction of growth, including posters and parallel sessions.
The natural thing to do would be to track the growth of the field. The question left is what do we gain by going against this natural trend? Does the conference have better papers this way? larger audience? more areas? better decision process, e.g. less sensitive to reviewer errors?
I’m in favor of larger STOC/FOCS but looking at the list of accepted papers, it does seem like that the average quality is lower than recent STOC/FOCS. Of course I could be wrong.
If this is indeed the case, I wouldn’t be surprised at the drop this year. This can also be an alarming call as it might be because people are drifting away from STOC/FOCS.
I agree with you that you could be wrong.
It’s no secret that the amount of (novelty)* required by a TCS paper is more than that required by a systems paper. This is not against the systems field, but rather an inherent difference between the nature of the two fields. I believe for TCS it is actually a question of whether (novelty) tracks the growth of the field. I also believe the answer is: it slowly does.
This makes the growth of TCS not necessarily/naturally proportional to the growth of the field. However, such growth in the systems field (or other CS fields) can be justified. To me, a large STOC/FOCS would raise a red flag: while it could be that all papers enjoy high quality (too good to be true), it most likely means many papers enjoy the opposite.
* I’m confident that I failed to come up with a more accurate word.
This makes the growth of TCS accepted papers not necessarily
I don’t have enough experience with systems papers to agree or disagree with that, and am not even sure how you could compare “novelty” between papers in different fields.
Furthermore, as far as I can tell the top systems conferences also did not grow so much even in recent years. (Looking at the first and last year available on http://www.cs.ucsb.edu/~almeroth/conf/stats/, SIGCOMM accepted 34 papers out of 50 submissions in 1983, and 38 papers out of 240 in 2013, SOSP accepted 22 papers out of 84 in 1995 and 28 papers out of 157 in 2011. Also just to get a sense of other venues – Nature magazine in July 17 1990 had 2 articles and 19 letters, while last week’s issue had 3 articles and 11 letters.)
I think the main issue at hand is that the physical constraints on people’s time and attention span remain about the same – there are only so many talks one person can go to, especially in areas other than their own, and only so many papers they can read – and that’s why conferences (or journals) are not growing with the field.
I agree that “novelty” is vague. I wanted to say that proving theorems is basically more difficult than doing, say, empirical experiments. Also, tools at one’s disposal could even be incapable of proving some theorems, in which case it could take years for the right tools to emerge.
As a rule I would hesitate to say that my field is more difficult or deeper than another field, especially one that I don’t know very well. I believe every field has its own types of difficulties, and I’m sure top systems work requires a lot of creativity and originality (though I really don’t have the expertise to comment on it). I am sure that a case could be made that coming up with a new idea for a system and implementing it is harder than proving a theorem, but of course this depends on which system we are talking about and which theorem.
Regardless, I don’t think this issue has anything to do with the number of papers in the conference. (Where if anything, systems conferences tend to be smaller and have a lower acceptance rate than TCS conferences.)
Thank you for such a great and organized work! I hope can elaborate how you run everything so the future chair can follow. By the way, I’m not sure if there will be a poster session or not. If so, please don’t let it be hijacked like the last STOC.