Wednesday, August 20, 2008

NSF Expeditions, Complexity

I'm glad to hear of the news that Sanjeev Arora's team at Princeton was one of the winners for the NSF Expeditions grants, working on the general theme of complexity. I think it shows that some of the public relations work our community has been doing, especially with the NSF, is paying off in concrete ways. I also think that more money for theory generally just has to be a good thing -- it's $10 million more for theory than there was before.

That being said, I'll express two concerns:

1) It's odd to see so much money for theory concentrated into such a small geographic area. I realize that was the nature of the Expeditions program, and I don't fault the proposal for it. It just strikes me as strange when the general budget for CS theory is so small to earmark such a large sum of money to this project. It feels like an over-concentration of resources in what's already a small community.

The solution to this, of course, is to get more money for the general CS theory program. And I'm sure a significant chunk of the Expeditions money will go to open DIMACS-style collaborations like workshops and other events, minimizing this concern.

2) I know it's just the nature of theory, but reading over the blurbs about the various funded Expeditions proposals, I can't help but notice that while the others seem to have some sort of statement of clear goals to take things in new directions ("hope to create a new field of computational sustainability", "It aims to create an "open" alternative to mobile ubiquitous computing and communication that can spur innovations, which will have a dramatic impact on the choices users will have in the way their data and information is computed, stored and communicated", "The project aims to develop tools and theories for molecular programming--such as programming languages and compilers--that will enable systematic design and implementation of technological and biotechnological applications that require information processing and decision-making to be embedded within and carried out by chemical processes."), the complexity grant will "hope to better understand the boundary between the tractable and the intractable" and "attack some of the deepest and hardest problems in computer science". Doesn't that sound, I don't know, just like business as usual? My concern is that it's probably important to the theory community long-term for this Expedition to have some major concrete success attributed to it at the end of the day. I have no doubt that good things will come out of this, just based on the people, who already do good work -- but will the output be the sort of things that in retrospect justify this investment?

6 comments:

James Lee said...

The project aims to develop tools and theories for molecular programming--such as programming languages and compilers--that will enable systematic design and implementation of technological and biotechnological applications that require information processing and decision-making to be embedded within and carried out by chemical processes.

"Doesn't that sound, I don't know, just like business as usual? My concern is that it's probably important to the biomolecular computing community long-term for this Expedition to have some major concrete success attributed to it at the end of the day."

Fortunately, they are not claiming they're going to cure cancer with their Expedition grant, and the Princeton folks are not claiming they're going to resolve P vs. NP.

I don't really see the difference, unless you think that their proposal (developing tools to do the things they've already been trying to do) is just more impressive. In that case, I would invite you to actually look at what they can do; what's the killer application that will come out in 5 years?

Anonymous said...

I have no absolutely no doubts this funding WILL have significant results. And I it will actually benefit all of TCS, because it will free more money to other people, as I am sure the priceton, IAS, Rutgers, and NYU people would have gotten their share anyway. In addition, the new center will surely be beneficial for numerous people in the area.

The problem of showing results at the end is there with every grant, small or large, and it boils down to PR work (assuming of course people do the technical work).

Do you think the other grants (e..g comp-bio one) will attain the goals they suggested? In most likelyhood, the entire community will start working on this right now (if they did not already).
Not to mention that the computational sustainability sounds like a PR balloon, compared to the sincerity of the theory proposal.

BTW, I think this goes to show that theory achievements are perceived as valuable and LASTING, as in core science compared to a matter of fashion.

Anonymous said...

The Neighbor's Grass Is Always Greener!

I am sure the argument about business as usual (the project would be carried it out anyway) holds true just as much for any of the other grants. In fact, they are probably already working on it, and also other people in their community...

I suspect it sounds more impressive and exotic when you are less familiar with it.

Robi.

Anonymous said...

When you give $10M to a dozen big-shot researchers at big-name universities, what do they do with the money? The same thing they've been doing all along.

NSF: welfare for the Ivy League.

OTOH, maybe it is best to only have a dozen research universities in the US.

Anonymous said...

Somebody sent me a link to this discussion. As Michael knows, I am also the chair for the Sigact committee for advancement of TCS, and our efforts are also leading to expansion of funding for theory in general.

Unfortunately (or fortunately) the trend towards giving larger grants seems irreversible, and there is no way for a specific field to "opt out" of such a cross-cutting program in exchange for increases in that field only.

If we hadn't got this grant, some other group in CS would have. It is not the same pot of money.

Our proposal was actually quite specific and broke up the larger complexity/algorithms agenda into short and medium term goalposts. (By the way, this is an important part of writing theory grants for cross-cutting programs; I'm happy to talk to anybody who plans to apply to such programs about and needs some tips).

"Business as usual"...

Believe me, we got a lot of grilling over this at all stages of this multistage process. There were almost no theory people on the final panel, and I am amazed that they were so receptive to our project. In fact, I think 3 out of 4 funded projects (out of 75 that entered the competition) are somewhat related to theory.


Sanjeev Arora

Michael Mitzenmacher said...

Sanjeev (and others) --

If I somehow mistakenly gave the impression that I somehow thought that the theory proposal was somehow unworthy, or even less worthy, than the other three, let me correct that impression. Nothing could be better for core theory than to have a theory expedition of this kind, focusing on fundamental research, and again, just based on the people involved on the co-PI list, I'm sure there will be great things coming out.

My specific point, regarding my reading of the blurbs, remains. Other blurbs were very specific about clear goals that were being aimed for, and/or what the "new direction" was. The theory blurb did not, and that stood out to me. I do disagree with commenters who do not acknowledge that point.

This seems to me to be a common hard problem in theory grants -- it's hard to describe what exactly will be done, and what the deliverables will be. I would be curious to hear more about the proposal, which I imagine contains more detail on this point, and/or provides other ways to convince others of the end value of the work.