Saturday, August 12, 2006

Funding

Hi, this is Claire Kenyon and I am the guest blogger while Lance is away. I have spent the last few weeks traveling and have heard many US academics express worries about possible pending money troubles. The precarious state of NSF funding is on everyone's mind. Like everybody else I know, I have sent a proposal in answer to NSF's last call for grant proposals in TCS, and I am now considering what the next step should be if it does not get funded. What alternative is there to NSF funding? I have asked this question to a few colleagues and have yet to hear a promising answer. However, I found the answer yesterday on my flight to Berkeley, reading the Kinsella best-seller, Confessions of a Shopaholic: "There are two solutions to money troubles. C.B., or M.M.M. -- Cut Back, or Make More Money." To know how to Cut Back, we can look at our less wealthy sister discipline, Math, for inspiration. For example, we could bypass conferences and submit our results directly to journals.

Comments page (Usual link does not work properly.)

12 comments:

  1. For example, we could bypass conferences and submit our results directly to journals.

    That's a terrible solution! Without conferences, there is no theoretical computer science.

    The NSF situation is every bit as desperate as everyone says. But I'm surprised that you don't discuss the M.M.M. (Make More Money) solution. Here's my proposal: some of us should get together and start an options-trading or software company whose sole purpose is to generate money for theoretical computer science. Perhaps we could all put in time for the company on a rotating basis. I think this would actually take up less time than the absurd business of writing and evaluating proposals.

    (Incidentally, to apply for a grant from this company, one would need to answer two questions: (1) "How much do you want?" (2) "What have you or your students proved recently?")

    But what if the company flops?, you ask. What if it doesn't make enough money? Isn't this a high-risk venture? Of course it is! The point is that continuing to rely on the NSF is an even higher-risk venture.

    ReplyDelete
  2. Your M.M.M. solution sounds a bit far fetched but I would join in if other people do it!!

    ReplyDelete
  3. I looked forward to Claire's blog stint, knowing that she'd write some interesting and provocative things, some of which I'd happen to totally disagree with.

    "...look at Math for inspiration..." That's really funny, Claire. (I also looked forward to Claire's sense of humor.) When I think of math and funding, I think desperation, not inspiration...

    I think bypassing conferences is an unworkable idea for TCS -- unlike math, we need the relativley quick turnaround -- but we could certainly consider doing things differently. How about a huge TCS conference every year, where say over half of the papers get in (garbage gets weeded out), and most everyone in the community comes? That could replace much of the FOCS/STOC/SODA/SPAA/PODC/Complexity/etc we have now; many of the smaller conferences might get reduced to local workshops. Instead of going to 3+ conferences a year in theory, you could cut back to 1-2, and not lose out all that much. Are there costs to such an approach? Obviously. But with scarce resources, it makes more sense to me to get everyone out for one big conference rather than three big ones and a half dozen smaller ones. More bang for the buck. I've gone to ISIT (Int'l Symposium on Information Theory) and INFOCOM, and yes, they're large, noisy, hard to deal with, and not all the work is so great. But it's a community focal point, people come, and it works reasonably well. Most other communities do this, why shouldn't we?

    ReplyDelete
  4. I think that both Claire's and Michael's suggestions miss the point:

    From Claire: For example, we could bypass conferences and submit our results directly to journals.

    From Michael: Instead of going to 3+ conferences a year in theory, you could cut back to 1-2, and not lose out all that much.

    If all we were talking about were travel funds then we could multiply the number of NSF grants by as much as a factor 10 and fund as much travel as NSF does now.

    The main issues are summer salary and grad students. Do we want to educate a next generation of theory practitioners? Do we want the insights of a new generation of brilliant researchers directed at theory problems? Do we want current researchers to continue their theory research or be lured away to other pursuits?

    ReplyDelete
  5. What Paul said. The main reason (I almost wrote "the only reason") to apply for research grants is to support graduate students, with summer salary a distant second. Travel and equipment are just noise.

    To follow up on Claire's suggestion... If we really want to follow math's lead, we would adopt the cultural stance that anyone who wants to do theoretical computer science for money is a heretic and an idiot, and they should be driven forthwith into the cogs of the nearest systems project and/or business school. Anyone who can be lured away from theory should be lured away from theory. Artists should be willing---no, grateful---to suffer for their art, forgoing pleasures of the flesh like meat and sleep in pursuit of Raw Truth.

    PhD students would be expected to support themselves entirely through teaching (except for a few rare fellowships, sysadmins, poker players, or comedy writers). Teaching positions would only be available to students with good research records, but the vast majority of teaching positions would be for remedial courses with hundreds of students, like "Information Literacy" or "Finding the Any Key".

    Conference travel would be paid by begging from the conference organizers, begging from the department, begging from the university, begging from parents, carpooling (from Cambridge to San Francisco), stuffing six people into a single hotel room, eating ramen noodles for a month, and leaving the grad students at home. As a result, travel would be much more rare, so all the real communication would happen online, via home pages, the ArXiv, mailing lists, and word of email. Electronic journals would finally, FINALLY take off.

    Once a year, there would be a huge meat-market conference with 1,000 attendees, the technical content of which consits of three-hour mini-conferences, most with less than 20 attendees. Except for the dozen or so richest (not best!) departments, new faculty would be recruited from the hordes of freshly-minted PhDs roaming the conference halls---the ones wearing ties/skirts and desperately foisting CVs on all passers-by---through a grueling day-long series of interviews, 15 minutes per candidate.

    Oh, yes! Let's do this, shall we?

    ReplyDelete
  6. I agree with Paul and Jeff that students/summer salary are a much higher order issue when it comes to funding. However, unlike travel, these issues are almost entirely outside our control. We don't set the rates for students or our salary. These are more or less binary decisions: we take a student, or don't, and we take summer salary, or don't. Travel is the rare part of the budget that is, more or less, under our individual control and the control of the community.

    Now, regarding the issues Paul and Jeff raise... (Aside: Jeff, I know you're busy with the baby and all, but please get back to writing viciously funny blog entries ASAP...)

    1) There is an argument out there that we produce too many Ph.D. students in CS (and TCS). I'm not saying I believe in the argument, but it certainly is an issue to keep in mind regarding funding and claiming that we want to get more students.
    2) As we've talked about before on this blog, I think we could do much better as a community publicizing our contributions to CS and science in general. This would go a long way for funding efforts.
    3) Personally, I must say that I find it bizarre that NSF seems out of touch with the rest of the world when it comes to the view of TCS. Departments have seemed to step up their hiring of theory people in recent years. Google, Yahoo, and Microsoft seem to love TCS grads come hiring time, and as far as I can tell, even prefer undergrads with strong theory backgrounds. Even in places like Wall Street (anyone heard of DE Shaw?), algorithmic skills are highly valued.

    Again, back to point 2, we need better PR, or Jeff's nightmare scenario seems all too close...

    ReplyDelete
  7. I agree with Paul and Jeff that students/summer salary are a much higher order issue when it comes to funding. However, unlike travel, these issues are almost entirely outside our control. We don't set the rates for students or our salary. These are more or less binary decisions: we take a student, or don't, and we take summer salary, or don't. Travel is the rare part of the budget that is, more or less, under our individual control and the control of the community.

    Now, regarding the issues Paul and Jeff raise... (Aside: Jeff, I know you're busy with the baby and all, but please get back to writing viciously funny blog entries ASAP...)

    1) There is an argument out there that we produce too many Ph.D. students in CS (and TCS). I'm not saying I believe in the argument, but it certainly is an issue to keep in mind regarding funding and claiming that we want to get more students.
    2) As we've talked about before on this blog, I think we could do much better as a community publicizing our contributions to CS and science in general. This would go a long way for funding efforts.
    3) Personally, I must say that I find it bizarre that NSF seems out of touch with the rest of the world when it comes to the view of TCS. Departments have seemed to step up their hiring of theory people in recent years. Google, Yahoo, and Microsoft seem to love TCS grads come hiring time, and as far as I can tell, even prefer undergrads with strong theory backgrounds. Even in places like Wall Street (anyone heard of DE Shaw?), algorithmic skills are highly valued.

    Again, back to point 2, we need better PR, or Jeff's nightmare scenario seems all too close...

    ReplyDelete
  8. Do not look at whether mathematicians are happy or not, but whether their field is thriving. It seems to me that Math research in the US is doing well.

    One very big conference once a year: that sounds quite unattractive to me. On the rare times on which I have attended FCRC, I have been frustrated by my inability to see people whom I knew were attending but whom I did not meet, just because of the size of the meeting.

    Student funding: sure, that's the number 1 issue, but what can we do about it?

    ReplyDelete
  9. I'm not sure what the figures are for math, but North American undergraduate enrollment in physics has been declining for forty years (two generations!).

    Fortunately, this has been compensated by an exponential increase in medical and biological research careers. But now, the era of growth in biomedical research funding has ended (about three years ago) and there is no immediate prospect that growth will resume.

    Just to be provocative, mightn't this stagnation of the science and technology ecosystem be blamed on ... NSF policies? On the grounds that, broadly speaking, technological societies require about 5 PhD-level engineers to support 1 PhD-level scientist.

    And therefore, can't we conclude, that over the past two decades the NSF should have been supporting engineering versus scientific research and education in roughly a 5-to-1 ratio, rather than the historical 1-to-5 ratio?

    I mention this because, at a PNNL workshop last week, the percentage of industrial customers for PNNL research services was reported as having declined to zero percent.

    Ecologically speaking, this is like a sea otter reserve that is inhabited by plenty of mature otters, but little kelp and little abalone. Under these conditions, the first thing that happens is that otter reproduction ceases. Later on, the population of otters crashes.

    ReplyDelete
  10. John,

    I'd hate to blame the NSF for the ecosystem problem. Ostensibly, the NSF sees its job as funding the scientist, particularly the "academic" research that wouldn't otherwise be funded but might have a large payoff, while industry/other government is supposed to be funding the engineering, practical side.

    I think the problems from the NSF come from other issues in the political winds. Overall budget issues, the desire to show immediate relevance (the follow-the-trend mentality), the difficulties in properly setting and modifying budgets in the face of entrenched interests (scientific and otherwise). But I don't see a lack of funding of engineering as being the problem.

    ReplyDelete
  11. I agree with you, Michael ... I was mainly trying to be provocative in suggesting that the science-to-engineering resource ratio at the NSF be inverted.

    I was serious, though, in pointing out that the overall US science and technology ecosystem is in poor shape, and that NSF budget stagnation is a symptom of it. And it seems to me that small-scale "patches" aren't likely to be very effective at fixing these large-scale problems.

    ReplyDelete
  12. "but whether their field is thriving. It seems to me that Math research in the US is doing well."

    Enrolments levels have yet to reach those of 1991 levels in spite of population growth and increasing participation in education. This is not by any means disastrous but neither is something to look for inspiration.

    ReplyDelete