Monday, September 21, 2009

Why You Shouldn't Not Go to FOCS

You can now register on-line for FOCS which includes the 50th celebration. Early registration deadline is October 1. Hotel rate good until October 9th or while supplies last. Be sure to register at the correct FOCS site http://www.cc.gatech.edu/focs2009 and not the fake focs2009.org.

There are some understandable reasons why you might not go to FOCS. It costs money. It takes time. You have a conflict or other responsibilities. You are just not into theory.

But recently I've started to hear a new excuse.
I don't have a paper at FOCS.
I've heard students say they are embarrassed to go to FOCS without a paper. Some say they don't think they are supposed to go without a paper. Some say it's "tradition" not to go to a conference unless you have a paper to present.

Hogwash.

I and many others went to every STOC and FOCS (and Complexity) as a grad student usually without a paper to give. It's a chance to find out the latest results, to discover the faces and personalities behind the names that you know. A place to talk to the current and future leaders of the theory community and give them a chance to know you. A chance to feel part of a large and strong theoretical computer science world beyond your own institution.

You should go to FOCS because you are a theorist and not just go only when you have the chance to toot your own horn.

36 comments:

  1. What if your advisor won't pay unless you have a paper?

    ReplyDelete
  2. Many people have complained that they are made to feel unwelcome at FOCS/STOC if they don't have a paper. This appears to be a different attitude to major conferences in other areas of CS and in other fields. The attitude seems to be "Who are you, why are you here and why should I even bother talking to you if you are not someone of any importance?" (importance is proven by having a FOCS/STOC paper)

    Perhaps this is related to Noam Nisan's latest blog post.

    ReplyDelete
  3. Some say it's "tradition" not to go to a conference unless you have a paper to present.

    Many people have complained that they are made to feel unwelcome at FOCS/STOC if they don't have a paper. This appears to be a different attitude to major conferences in other areas of CS and in other fields.

    TCS seems to have some strange traditions and attitudes. I thought academic conferences were, among other things, a place to facilitate the discussion of ideas and the exchange of knowledge. If you have something intelligent and relevant to say/ask, why should you be ignored? These meetings should not be set up like country clubs.

    ReplyDelete
  4. "Who are you, why are you here and why should I even bother talking to you if you are not someone of any importance?"

    It's been a few years since I was a grad student, but I walked away from STOC/FOCS feeling the same way. This did not happen at other conferences which I attended also without a paper.

    ReplyDelete
  5. I don't know about US, but in many places in Europe it is hard to get funding for passive participation. Some grants even include a clause which in effect says "active participation only"

    ReplyDelete
  6. Time has changed. Lance.
    There is really no point to
    go to FOCS if you do not
    have a paper. If you really
    have money to burn,
    go to some focus workshop.

    ReplyDelete
  7. How many people generally turn up to the biggest TCS conferences?

    ReplyDelete
  8. This is not a new excuse--it's a very old excuse. If FOCS wanted to include the community more, they would have lower registration (non-country club lunches, etc) and a rump session.

    Right now it is too expensive and there is not a clear way for everyone to participate.

    ReplyDelete
  9. Why can't FOCS/STOC have reasonable registration fees under a variety of categories, like they do at the annual AMS meeting?

    http://www.ams.org/amsmtgs/2124_regfees.html

    ReplyDelete
  10. I have to say, I read this, and also found myself disagreeing with the sentiment -- but I think it's important for discussion.

    I think as a starting point we should recognize that other communities do have conferences where much of the community attends. For some conferences, like ISIT for information theory, the structure is that the bar is relatively low and the number of papers is large; for others, like SIGCOMM, the structure is that the bar is very high and the number of papers small, but the community makes a concerted effort to attend.

    The theory community does not have such a venue, and asking why is a reasonable question. There seem to be multiple reasons:

    1) Money : cost is an issue for many researchers at all levels, from senior to student. (Networking people, on the whole, seem to have more money for, say, a trip to Barcelona for SIGCOMM.)

    2) A fractured community : the theory community seems much more seriously split into subcommunities than others. I don't know why that's the case. (Possibly it's massive specialization in theory, but I'm not sure that's really the case.)

    3) Too many supposed "must" conferences : Even having two "major" conferences in FOCS/STOC (never mind a 3rd in SODA) I think provides a strong negative impetus for attending one of these conferences when you don't have a paper. Do others see this as a problem for building a community?

    4) Attitude : Clearly, lots of anonymous people think there's an attitude problem. Perhaps other modes of participation -- rump sessions, open problem sessions -- to open up the conferences is something to be encouraged.

    Lots to think about in each of these areas.

    ReplyDelete
  11. Hi Lance. From the comments above you can understand the point is if the conference worths the money people pay, specially for students when their adviser or department is not paying for them. So you can not just separate it.

    ReplyDelete
  12. Cost is something that comes up at every business meeting, as well as in these anonymous comments. The situation never improves, I think because the decision is that catered lunches and nice venues are more important than cost.

    It seems that a radical change would be needed to lower costs significantly. Back in the day, everyone may have gone to FOCS and STOC, even without a paper, but they were also staying at much cheaper hotels back then. Catered lunches might be a plus for grad students who can't afford restaurant choices made by their seniors.

    Grad students: You can stay at other hotels, and save up to half off on that portion of the conference. The conference organizers don't promote this option because they have to sell their block, but use travelocity. Also, personal expenses are tax deductable even if you can't get reimbursed.

    ReplyDelete
  13. Re. comment by Michael Mitzenmacher:


    3) Too many supposed "must" conferences : Even having two "major" conferences in FOCS/STOC (never mind a 3rd in SODA) I think provides a strong negative impetus for attending one of these conferences when you don't have a paper. Do others see this as a problem for building a community?


    I am surprised by this remark coming from you, Michael. What happened to the power of two choices :-)?

    Seriously, having two copies of very similar conferences is good in the sense that as long as one can go to one of these depending on various factors (teaching schedule, proximity, (ahem) having a paper or not having one, possibility of visiting collaborators in the vicinity of the conference, etc.), I think the community will stay well-connected.

    ReplyDelete
  14. Economic considerations have not yet been mentioned, and yet their influence on grad students undoubtedly is strong.

    Academic jobs are scarce and likely to remain so for awhile; students who have no conference-quality results (yet) obviously are less motivated to attend.

    Only a contrarian investor would argue otherwise ... on the grounds that the best time to invest (in an academic career) is at the market bottom.

    How many people (students or faculty) take the contrarian view, perceiving that a boom in complexity theory (in both the private and public sectors) is just around the corner?

    I ask this because my own views are decidedly contrarian. :)

    ReplyDelete
  15. STOC/FOCS once were the focal conferences of TCS. They forsook this role long ago when they chose not to keep up with growth of the field.

    Today SODA is much closer to being the central conference of the field both in attendance and range than STOC/FOCS have been for quite a while.

    If STOC were to triple the number of accepted papers, add satellite workshops and rump sessions and move its location to a university campus in the summer it could quickly become central again. Sadly it is unlikely to happen. There is too much inertia in the system.

    ReplyDelete
  16. I'd very much like to go to STOC or FOCS regularly. Many people have mentioned time/money as reasons why this doesn't happen. That's true, but the real question is the prioritization, since we all use that time and money and presumably go elsewhere.

    For me, STOC/FOCS lose out in prioritization because

    * I *am* compartmentalized. I work on a range of problems that more often show up in SODA, SoCG, and even conferences in databases and data mining. So I'm likely to publish in these venues, under the premise that the people who care about these problems will also show up there.

    * If I just want to know about the latest results, it's more time effective to just browse the papers. There's no need to go to a conference just to hear a talk, unless there are many such talks, in which case, see above.

    * I miss the hallway conversations etc, but probably the most useful bit of gossip is of the form "this group has already improved the results in this talk blah blah": that can't be easily replaced online. But again, that in itself is not enough to justify the travel.

    * I'd obviously attend STOC/FOCS if/when I publish there. but apart from that, it's hard to make the case to trudge over, unless (as some suggested), the scope of non-talk activities (tutorials, workshops etc) is expanded dramatically. This is something that is common place at conferences in databases and machine learning, for example. I think I'd definitely consider attending STOC/FOCS if I could (in addition to talks) learn about the latest in a range of topics of active interest (AGT, Fourier methods, additive combinatorics (like Luca's tutorial) etc etc). But I'd need more than one.

    ReplyDelete
  17. D. Sivakumar --

    I'm surprised at your misinterpretation of the power of two choices. :)

    The power of choice is good if you want to spread load among multiple choices. Indeed, that's precisely what having FOCS/STOC/SODA is doing -- people are load-balancing among the 3 conferences based on the criterion of whether they have a paper there. (Indeed, the point in my post was I believe people think why should I go to FOCS if I don't have a paper there; I'll get my STOC paper ready and go to that instead...) You point out this flexibility may increase the absolute number of people attending these conferences (which I am skeptical is actually true), but it clearly reduces the maximum load you'd see at any one conference -- that is, we lose the "everyone shows up" mindset that seemed to be what Lance was promoting, and does occur in other communities with fewer supposed "must-attend" conferences.

    ReplyDelete
  18. A large fraction of FOCS/STOC attendees are students (30-50%). The vast majority of them do not have papers. Student registration of $190 is cheap and includes a lot of food. (It basically covers the incremental cost.)

    The conferences are not only an opportunity to get to learn more from/about the leaders in the field, they are also an opportunity to meet and get to know other students who will be your peers in years to come. One should not be shy. Meet as many people as you can. Wear your badge prominently so people have a point of reference to start up a conversation. Spend a little time hanging out at your advisor's shoulder so they will introduce you around. (Don't overdo it, though.) There will be conversations that are way over your head - don't worry about that.

    ReplyDelete
  19. Paul --

    I agree that students should make every effort to go to conferences -- even without a paper -- when possible, and it's a great opportunity to learn.

    For grad students, ostensibly time isn't the limiting factor -- hopefully you're not too busy at this stage to go to a conference -- and I imagine the major issue is cost. Definitely go to local conferences (heck, even ones outside your main interest are probably worth attending), where hotel/flight costs don't come into play. For more distant conferences... maybe we really need an organized way for grad students at one place to put grad students from another up on their couch or something. Or at least a way to get 4 grad students to share a room. Hotel costs over 3 days can often be the main expense of a conference.

    ReplyDelete
  20. Most of the cost arguments for graduate students are too short-sighted. Airfares are historically cheap these days. I just checked and saw that the fares to Atlanta from Seattle are $220 right now. It is possible to share rooms at low marginal cost; a room at the conference rate is little more than $200 per person for 3 nights and alternatives are even cheaper.

    If you think about the investment of your time in graduate school, the dollars you would have to pay even if you are not supported is an investment in yourself that is a small fraction of what you have invested already just for being a grad student during the month of the conference.

    In the early 1980's as a grad student earning less than half of typical grad student stipends these days, it cost me over $550 out-of-pocket just for airfare to attend and it was well worth it.

    P.S. One other tip for grad students: If you show the initiative and end up going to a conference on your own, your advisor is often willing to pitch in some money for things like registration after the fact.

    ReplyDelete
  21. As an outsider, I wonder whether this is a naive question: Why not have a joint CS meeting, like the math people do each year? The cost to attend should be reduced, the atmosphere would be more inclusive, and inter-group interaction would be convenient and encouraged.

    ReplyDelete
  22. A large fraction of FOCS/STOC attendees are students (30-50%). The vast majority of them do not have papers. Student registration of $190 is cheap and includes a lot of food. (It basically covers the incremental cost.)

    Do we have recent figures for this? The last two STOC/FOCS I went to seemed very sparsely attended, nowhere near to 40-50% of attendees without a paper.

    ReplyDelete
  23. Michael (re. comment 18): if there is one premier event each year and if for some reason I can't go to it (scheduling, personal reasons, location, cost, etc.), I go two years without being at the main event - so load balancing helps minimize the maximum (or average) gap of attendance for authors, so there will be more overall interaction.

    ReplyDelete
  24. I just want to point out that sharing a room at a conference with others is a bit of an issue if you are a woman... Also, I have found that unless you are super-social you tend to be completely ignored at conferences, especially if you don't have a paper.

    ReplyDelete
  25. Attitude : Clearly, lots of anonymous people think there's an attitude problem.

    I'll go on record, what the hell.

    I haven't been to either STOC or FOCS, and I have only good things to say about how the STOC 2009 program committee dealt with my submission to them, but I did notice something about the conversation at the Barriers Workshop that was very different from anything I'd seen elsewhere.

    Instead of "What's your name? What are you working on?" which is the standard intro protocol at distributed computing or molecular computing conferences, there were a lot of people at Barriers who instead asked me, "What's your name? What school are you from?" And, when it was clear I wasn't at an Ivy League school, the conversation was literally over. There were both students and "famous faculty" who were gracious and actively interested in having conversations with me, but maybe half the people I met came off as more prestige-oriented than scientifically curious.

    I interpreted it as a circle-the-wagons symptom of a desultory field, where people are afraid there will be almost no jobs in pure complexity, so being cliquish and taking care of your own first is the way to go. Once I returned home, some friends asked me what I learned at Barriers. I responded by saying, "It solidified my analysis that I should apply complexity theory, instead of 'doing' complexity theory."

    I'm going to repeat something, because I think it's important: I haven't encountered that you're-not-important-enough-to-talk-to phenomenon at any other computer science meeting I've been at. I know it's not just me, either, because I talked to a few other students at Barriers. One said, "Theory people just aren't very nice." Another said, "Aaron, there's just a Princeton clique. If you don't have something to offer them, they don't want anything to do with you." Both those statements are stronger than anything I'd be willing to make myself, but, well, wow. Someone needs to rewrite the marketing strategy, you know what I mean?

    ReplyDelete
  26. D. Sivakumar --

    so load balancing helps minimize the maximum (or average) gap of attendance for authors, so there will be more overall interaction.

    If you extend this to the logical conclusion -- we should have a conference every day -- I think you'll find your conclusion (more overall interaction) just doesn't hold up.

    There are, I think, two reasons for this. First, interactions probably scale (up to reasonable sizes) proportionally to the square of the attendees at the conference. (Assuming two random people at the conference want to interact with probability p, and the number of attendees n is such that np is not a bottleneck to communication.) So two conferences with 100 people probably gives about as much opportunity for interaction as one conference with 150. (Note that when n is large -- when your conference gets "too big" -- the communication doesn't scale, and you may want to split to smaller conferences.)

    Second, and this is what I hinted at but didn't explicitly say before, there's a feedback effect. Why does everyone go to SIGCOMM? Because everyone else is going to SIGCOMM. (Indeed, I'm always surprised that people expect me to be going to SIGCOMM when I don't have a paper -- and this expectation is reaching the point where soon I think I'll end up going...) You give lots of reasons why people don't go to conferences -- teaching, distance to travel, etc. In my experience, in communities where there is one "big" conference, people reprioritize and go to the meeting -- because it's important! Again, more overall interaction.

    ReplyDelete
  27. To Aaron: As a student at a top-10 school, I can tell you that being in an Ivy League doesn't help you much either :). Were you in an Ivy League school, the next question would be: "Who is your advisor?", and if your answer were not satisfactory, the conversation would be over again !

    ReplyDelete
  28. To MM, re larger conferences leading to more interactions: my experience is the opposite. In a small workshop (say 20-30 people) I'm likely to interact with a large fraction of the participants. For a larger conference with 100-150 or more people, I'm likely to either hang out with a small clique of friends (if I already know people there) or get lost in the crowd and not interact with anyone (if I don't).

    ReplyDelete
  29. Do we have recent figures for this?

    At FOCS 2008 there were 122 students out of a total of 270 paid attendees.

    ReplyDelete
  30. David --

    I agree -- at specialized workshops, there is more communication, because it's specialized. That wasn't what we were talking about here -- we were talking about FOCS/STOC style flagship conferences for theory, where you're not going to get that kind of specialization. So we aren't in disagreement -- you're just talking about a different beast.

    ReplyDelete
  31. Were you in an Ivy League school, the next question would be: "Who is your advisor?", and if your answer were not satisfactory, the conversation would be over again !

    Well, that's just intellectual suicide. It's so ironic. I've read a lot of online slides, and maybe my favorite set ever is "Depth through Breadth" by Wigderson, where he makes the point that you get stronger in your own area by gaining understanding of other areas. So it's not as though "no one in the upper echelons" understands this.

    ReplyDelete
  32. Aaron: The other possibility is that complexity theorists don't know how to make conversation.

    ReplyDelete
  33. I've read a lot of online slides, and maybe my favorite set ever is "Depth through Breadth" by Wigderson...

    Interestingly enough, his final slide says: "FOCS/STOC: well attended ... open, inclusive ... tolerant to new (weird?) ideas ... student friendly..."

    ReplyDelete
  34. Paul Beame: "Most of the cost arguments for graduate students are too short-sighted... If you think about the investment of your time in graduate school, the dollars you would have to pay even if you are not supported is an investment in yourself that is a small fraction of what you have invested already just for being a grad student during the month of the conference."

    Grad students aren't in it for the money. They correctly do not see grad school as an investment in themselves, since there are no jobs at the other end. Scott Aaronson's point of view, that it is more like entering the Olympics, is more accurate. Grad students do what they love, hopefully.

    ReplyDelete
  35. Grad students do what they love, hopefully.

    Of course. I was only trying to broaden slightly the narrow economic terms on which the cost complaints were made. Lance made a lot of non-economic arguments already and I could add more there too. The non-economic case is at least as strong as the economic one.

    ReplyDelete
  36. EuropeanGradStudent12:31 PM, September 22, 2009

    Just to let you known that in many European departments a grad student may get a remboursement if and only if she/he has an accepted paper.

    If you plan to attend a major conference in the US, you'll have to afford the attendance fee, a transoceanic flight and 4+ days accommodation, and this can be done - with your personal funds - once or twice in the whole grad school.

    At least, please decrease the fees, by choosing cheaper venues!

    ReplyDelete