Monday, May 03, 2010

The Computer Science Ecosystem

Inspired a bit by some inspirational, high-level talks I've seen the last few days (see Matt's take on Ed Lazowska's talk at Harvard here), I've been thinking about what I'll call the computer science ecosystem.  The inspirational talks often refer a lot to the grand sweeping visions -- science will all be done on the cloud, computerized automobile systems will make driving safer and more efficient, etc.  These generally sound amazing, although to some they can also sound a bit out there.  (Robobees and Sean Hannity, anyone?) On the other hand, there's a lot of basic work down in the trenches creating the building blocks that eventually make the sweeping visions possible.  While some of that work is explicitly done as part of these bigger projects, a lot of scientists are working on their own little pieces of the puzzle without specific regard to the big picture.  And a lot of good work gets done that way too.

So what is the right sort of "research ecosystem" we should have, that balances between BIG and small projects, between individual researchers pursuing their goals and larger groups, between researchers focused on "core areas and problems" and researchers focused on interdisciplinary problems?  How do we promote and build such an ecosystem that lets the various types of research flourish, hopefully in ways that the sum is greater than the parts?

I don't pretend that I have the answers to these important questions.  They're certainly thought about over at the NSF, which decides how much money to distribute among small, medium, and large proposals, and how much to give to different programs, some of which specifically aim to promote interdisciplinary research.  And discussions about these questions have been (and I assume still are) a significant issue for the ACM-SIGACT Committee for the Advancement of Theoretical Computer Science, which gives feedback to the NSF on such matters.

Certainly a concern that always springs to mind is that the funding model for CS will come to be dominated by big projects, leaving little room for the individual researcher or even small groups more common in theory.  But in terms of increasing the funding pie overall for CS, the big project ideas are often much more persuasive, which seems to be the push taken by the Computing Community Consortium.  I don't know currently what the dollar spend looks like from NSF or the other funding agencies;  right now it still feels to me that we have a suitably diverse research ecosystem, but perhaps my view is limited.  


Daniel Lemire said...

My general feeling is that funding agencies don't like diversity very much. They tend to proceed in a top-down way: setting up goals, and then executing them. They have this model of the ideal research project, down this its attributes.

But then, let's remember that while Gregori Perelman had a salary, he did not, as far as I can tell, have (or need) a grant.

Anonymous said...

What do you mean by that statement about Perelman?

Perelman was a postdoc/visitor in the US for several years. He saved his money and used it to support himself back in Russia, a much cheaper country than the US.

Are you suggesting that theory people should save their money and do without salaries and grants?

Daniel Lemire said...


Are you suggesting that theory people should save their money and do without salaries and grants?

Yes, I'm actually suggesting all researchers follow the path of the unabomber and go live in shacks. In fact, I'm sending back all grant money I ever got, and I am renouncing my salary as of right now.

Come on!

No. What I am saying is that if Perelman had been busy applying for million-dollar grants to do this and that, he might have done different work. Would it have been better or worse? I don't know. But I bet it would have been different.

That's all I'm saying. (And no, I don't live in a shack.)

Anonymous said...

"Perelman was a postdoc/visitor in the US for several years. He saved his money and used it to support himself back in Russia, a much cheaper country than the US."

I am a postdoc in US for like 2 years. I do not think that my savings made in US, would last for a long time in India :(

Anonymous said...

If you look at Perelman's arxiv papers on the poincare conjecture, at least one acknowledges that he used savings from paid visits/positions in the US to support himself back in Russia.

Anyway, doesn't sound like he has a high budget.

D Lemire makes a good point about the time spend writing grants.

Anonymous CS Prof said...

You mention the question of how NSF should allocate a fixed amount of funding among small, medium, and large/center-scale projects.

My opinions: NSF is spending a lot of money on large/center-scale projects that I suspect might be better spent on smaller grants.

I believe the reviewing process for large/center-scale projects is not great. You get the same 12-15 pages to describe your ideas for a ten-million dollar center as for a several-hundred-thousand small grant. Reviewing for center-scale grants seems to depend a lot more on non-substantive factors, e.g., can you spin the idea really well, are you a really good grant-writer (rather than a good scientist), are the people involved famous, and so on. The reviews on a $10M center-scale grant is not 20x as rigorous as the reviews on a $500K small grant; if anything, the reviews on the small grant tend to be more rigorous on the technical details, in my experience. And for large centers with many participants across many institutions, conflict-of-interest considerations not infrequently rule out most of the best experts in the field, leaving center review panels with people of lesser quality than you'd see on review panels for small grants.

My sense is that if NSF spends $10 million on a single center, odds are that a good fraction of the work will be of inferior quality to what they could have gotten if they had funded 20 $500K grants. Why? My guess: centers get a big pot of money, then the center redistributes that money internally among their own members, and the methods that centers use to allocate money internally are less rigorous and less merit-based than the methods that NSF uses to review proposals for small grants. My impression is that it is not unusual for inter-personal and inter-institutional politics to play a non-trivial role in center internal allocation processes. Conflicts of interest that would never be tolerated when reviewing a small grant proposal appear to be routine in many centers' internal funding process.

Don't get me wrong. I've been on centers that I think have been very positive and where I think NSF's investment led to tremendous returns. However I've also seen some more dubious cases. My impression is that the variation in quality among center-scale grants is noticeably higher.

I've sometimes wondered why the NSF spends so much money on large centers, when overall funding rates are so low (are they still in the single digits?). I wonder if centers have the benefit that they're easy to explain to Congress in a sound bite, whereas it's harder to explain 20 disparate unrelated small grants in a sound bite. Perhaps I'm being too cynical.... Alternatively, maybe the issue is that support for centers comes from a different pot of money, and it's not easy to shift money around between categories?

I realize this comment may sound like sour grapes. It's not intended that way. I've been generously funded through both small and large grants, and I'm extraordinarily grateful for the level of support the NSF has provided. The NSF is a fantastic institution. I've probably received more than my fair share of funding from large and center-scale efforts, so this is not bitterness about the level of funding I've personally received. My comments are intended in the spirit of examining what allocation of money leads to the best science overall, given a fixed amount of money.

Doc Terror said...

Every time I go to my local Uni library and walk through the stacks I shake my head. There are practically three shelf segments (150 feet of shelf) stuffed full of conference proceedings on IP QoS, yet, nothing practical has come out of this research for 30 years.

I can think of many areas in computing that could transform industries, transform society, that are "redlined" so far as CS funding is concerned. I worked on a project that transformed scientific publishing that struggled to put together $250K of "soft money" financing; in the same building there was a project that burned $2M of funding on a web site that never found an audience and a stream of mediocre conference papers.

For instance, semantic technology is starting to break into commercialization and it looks like Europe, for once, might have the jump on the US. At an "afterparty" to the International Semantic Web Conference that was held in Washington, DC last year, I didn't encounter a single American computer scientist: I met entrepreneurs, physicists and unemployed digital librarians, but nobody in "CS".

I see that computer science is where physics was in 1968. I sit in the back at CS "job talks" at my local Uni with a single minded focus on the commercialization of technology. I see PhD students who are looking for permanent jobs in academia or who are looking for research jobs at just a handful of large paternal companies (Google, Yahoo, Microsoft, IBM.)

That's the way it was in physics when the need for professors was rapidly expanding, and when large industrial labs offered desirable jobs in research. By 1970, this situation broke, and physics PhDs found themselves driving cabs in NYC. The job market in physics has been moribund ever since; today the only reliable way get a professorship in physics is to (i) have a parent who is a professor and/or (ii) be an affirmative action case -- preferably both.

The fact that the postdoc institution is diffusing from rapidly physics into CS is slowing the realization of the pain in computer science, but ten years from now, when a bunch of people realize they're almost 40 and have built no equity in their careers, it will start to dawn on people.

Anonymous said...

How much of this tradeoff is determined by the NSF and how much by CS departments. I imagine many CS departments would not be interested in hiring a Perelman, since he doesn't bring in overhead.

Antonio B Duarte Jr said...

We really apreciate your bolg. Team of Curso de Informtica Online.