Kolata on Funding Research
“Grant money goes to projects unlikely to break much ground”-Kolata
Gina Kolata is a science writer for The New York Times. She is a wonderful journalist who understands science and how to make her stories come to life. I have had the honor of being the “subject” of several of her articles over the years and, while at Princeton, was one of her background sources for a few of her pieces. Gina coverage of the first DNA conference that Eric Baum and I organized years ago helped make it a success. Now that I am in Atlanta and she lives in Princeton and commutes to New York, I seem to have fallen off her rolodex. Oh well.
Today I want to point out a important article that just appeared today in the Sunday Times–above the fold. Gina’s article is entitled, Playing It Safe in Cancer Research. I strongly recommend that you read her article. The subtitle is a perfect summary of the article: “Grant Money Goes to Projects Unlikely To Break Much Ground.”
Conventional Wisdom
I believe that everything Gina says about funding for cancer research applies also to funding in the rest of science, and specifically in the theory of computing. The point of her article is that “wild” ideas that could change the landscape are usually not funded. As a result, some researchers do not even submit their potentially-breakthrough ideas to the funding system.
Eileen Jaffe is a senior scientist who was turned down–rejected out of hand–by NIH without even a panel review. Gina gives details on Jaffe’s idea, but to a layperson the idea sounds terrific. One of the reasons given for the rejection was that she had no preliminary results: Jaffe said:
“Of course I don’t. I need the grant money to get them.”
I want to be clear, cancer research is much more important than funding theory, perhaps more important than anything else. The point I am making is their funding system suffers from the “fear” of taking chances, and as a result ground-breaking research is held back. It’s all about fighting conventional wisdom. If everyone thinks it is obvious that factoring is hard, then who would get money to work factoring algorithms? If everyone thinks it is obvious that P is not equal to NP, then who would get funded to work on proving P=NP? And so on.
Our funding support–which for us means NSF–must be willing to take chances. NSF must be willing to fund high risk, high payoff research so our ground-breaking efforts are not held back. Our community must be willing to embrace those who have different views, and to support them with money and in other ways.
If we do not do this, I think the field of theory will not continue to flourish, and will not make the breakthroughs that it is capable of doing.
Open Problems
Please take a look at her great article. Also think of what we can do so that risky ideas can be tried out.
I don’t think it is nearly so big a problem in theory. I wouldn’t support a grant application to try to prove P = NP, or for finding an efficient classical factoring algorithm, unless there were some strong ideas in the proposal. But in TCS, strong ideas are essentially “preliminary results.” In biology, you might need extra money to purchase equipment, pay for lab staff, etc. Where this money comes from is a Catch-22. In TCS, you don’t need to $500 thousand to start to think seriously about a problem. Although it isn’t supposed to work this way, I think grants in TCS are more likely to be awarded for good previous work, and actually used for other purposes entirely. (It would be nice if this were official policy.)
I am also skeptical of the article. There are surely two sides to this story, but the reporter only gives us one of them.
I was not going to add this, but I will now. I have done well over the years with my own NSF funding. However, I remember two times when I tried to go outside the “box”. Both times I got zero dollars. Nothing. The first was picked up later by DARPA, which at the time was less risk averse than NSF.
The second, was an approach to the factoring problem. I know its a long shot, but I still like the idea. I have and will post some more on my ideas about factoring. However, the comments sounded identical to those of Jaffe in the Gina’s article. The theme was: “Where are your partial results?” Partial results? Factoring is kind of 0/1. Either you can factor or you cannot. What partial results could I possibly have?
Thanks for the link – very interesting! However, as deplorable as the situation is, I wonder how much better it could be implemented in reality. Though innovation and creativity might increase, it seems that the amount of people trying to “game” the system will correspondingly rise as well. I wish it were different, but accepting riskier ideas might encourage researchers to write fantastic-sounding proposals without fully believing in the outcome. If most new medicines fail with today’s stringent policies, what if those policies were relaxed? It’s not so clear.
I like the idea of creating these “challenge funds”, which support high-risk ventures. But how to filter out all the chaff? Also, I think this climate might encourage scientists to consider unorthodox, but original ways of conducting novel research without having to spend millions of dollars on equipment. I believe that sometimes, science is carried out by first throwing money at things, before coming up with an idea.
I agree with Jon above – with TCS, it’s a little easier. It usually doesn’t require millions of dollars to get some preliminary results of some kind.
Good link, good article, good post.
I strongly disagree with the previous commentor: you would not fund p?=np? I hope you are not a CS theory researcher, because if you are, then you are admitting that the problem that is considered and promoted as the most important question in CS theory, is not so important. If you are a CS theoretician, then you have a non-mainstream view of theory. (You must be having trouble securing funding, no?)
Sometimes there aren’t two sides to the story, Jon.
In TCS, one might not need the money to build particle accelerators with, but one does need the money to—surprise, surprise–ATTRACT STUDENTS with!!!
I truly truly truly believe that many deep questions are attackable but for the lack of courage and leadership. AKS were not afraid to work on the problem and were supported by their funding agencies. We have to set our eyes on the big prizes, otherwise we will be left far behind. This is what theory is about.
I once read that Fortnow also would not advocate for P/NP research. What are these giants smoking?
What Jon means is that he wouldn’t fund research with the explicit goal of showing that P=NP, presumably since he believes that the opposite is true, hence all research should attempt to show that P!=NP instead. Which was, I believe, Dick’s point: it’s hard to justify funding research that goes in a direction that many people believe is hopeless, but that doesn’t mean that this research isn’t important.
Richard —
I think the balance between risky and regular research is a difficult and challenging issue. In a more perfect world there would be more available funds so that it would make sense to take a percentage and say, “This money is set aside for really high-risk, out there projects.” Indeed, NSF does have funds of this sort now as I understand. But it’s hard to set aside money for that when there are more good projects (and researchers) than funding, which is the state I think CS theory is in and has been in for some time.
In fact I’d rather see NSF have a more explicit “Google-like” policy, where they’d say, “We’re funding your project, but we expect xx% of your time to be risky, forward-thinking stuff.” I actually think this is implicit in CS theory funding, where most people can’t help but think of stuff that’s more out there — it’s how many people in the field think. But it might be nice if it was more clearly and formally recognized that this is a good and desirable thing as part of the standard grant process.
Deciding whether or not to fund a particular proposal is not a judgement based only on whether the proposed problem is important. If the NSF funded every proposal to solve P =? NP, then every proposal it got would be for that problem. And then what?
Yes there are two sides of the story in this case. The people rejecting proposals are themselves researchers.
Regarding attracting students, you should not be attracting students with the promise of solving P =? NP. And if you have a problem for which you do not think partial results are even possible (I am not sure that factoring is an example, but it might be), then I also think that is not a good problem for students. This is very high risk research. The possibly high reward may be worth taxpayer dollars, but it is absolutely not fair to risk a student’s career on it. A student’s career cannot be a 0/1 thing. Students can work on these problems, but they need to work on other problems, too, and that is how you can fund them.
Right now a low-risk, low-reward proposal will not be funded, and neither will a high-risk, high-reward proposal (without intermediate results and a clear plan). There is a bias toward lower risk, high reward research, and that is how it should be.
I truly truly truly believe that many deep questions are attackable but for the lack of courage and leadership.
Then you should definitely definitely definitely provide some of that courage and leadership.
I agree with you that there’s too much timidity in the theoretical CS community about various famous problems. The way to fix that is to propose some methods of attack and get people excited (or to solve the problems yourself if you prefer).
There is very good discussion of this story at Respectful Insolence by an oncologist. Look at the entry linked and the especially the next one. Most interesting was the fact that several of the examples Kolata cited as innovative work that would not be funded by the government actually WERE funded by the government. I don’t know if this was selective memory or what.
Speaking for myself as an individual I’ll go as far as making this argument: Federal science funding should be quite conservative. It should NOT try to pick innovative research to fund, it should rather spend money on research it thinks has a decent chance will lead to something. Why? Because there is no evidence that Federal agencies are any good at all at picking far out research that will be successful. That means that this money is very, very likely to be wasted. If you want to argue that by funding this we are taking a long shot at a big win I’ll say this is the same reasoning used to justify poor people spending on lotteries and I am against that also.
If you want to take high risk gambles then philanthropical organizations or private moneys should be used.
I don’t want every wacko, alternative medicine producer and who knows what else consuming time and resources. And they would! You will have homeopathists wanting money, intelligent design studies will be proposed with good legislative backing. Global warming “alternatives” that have as much to do with science as fundamentalist Christianity. The effort it would take to deal with this junk could cause us to miss important results that would emerge from the solid well grounded research. I, with as much evidence as the other way, think this would be the more likely outcome – less breakthroughs and solid science undermined by deliberate anti-science groups that are always working to make government funded research look bad. In fact fact the more I think of it the more the downside of this idea looks horrific. You will end up with new Proxmire’s with good public support slashing things.