Sunday, December 05, 2010

Ideas for fixing science, contn.

Shooting the bullshit with a friend over beer, we started talking about a couple of articles that came out recently, including this one about productivity and money and this one about doubling the NSF budget, among others.

My friend was saying it can be wasteful to spend too much on basic research.

I was arguing that lots of basic research has applications later, like GFP and RNAi. There are plenty of examples of initially esoteric findings that turned out to be incredibly useful.

We were saying how it's important to fund both a sizable chunk of applied science and a healthy chunk of purely basic research, but we're not exactly sure what the best ratios would be.

Then we were talking about how there's no incentive to get a lot of bang for your buck. I've written before about how the average Cell paper costs about a million dollars or more, when you add up salaries and expenses to do the amount of work that is expected for "a complete story". Somewhere along the line (and I saw Drugmonkey mentioned this not long ago), quality got overrun by quantity. Now it seems like the only way to get high impact papers is to have a lot of money to spend, and the only way to get more money is to get high impact papers.

And thus, a cycle of nastiness was born.

But hey, that's NIH. Maybe NSF is not so bad (yet)? Currently, the NSF budget doesn't really cost much money, relative to other government spending.

On the other hand, just pumping money into it isn't going to solve everything, and if the NIH doubling is an example, it's likely to create a whole buncha additional problems that NSF hasn't solved before.

I said again that I think the problem is partly how research careers select for people who are okay with not trying to predict the future too far in advance. We'd all go crazy if our experiments didn't always turn out like we expected them to. Instead, happy researchers are the ones who are okay with that kind of uncertainty. We make plans, but we don't feel too compelled to stick to them if they don't make sense.

However, if you're going to run a government funding agency, and that's going to affect workforce issues, you kinda hafta be the sorta person who likes a PLAN. You really have to think about the consequences of your plans when it's going to affect entire generations of trainees not being able to get jobs fifteen years from now.

I don't know what's going on with the people at the top of the funding agencies, but Jeremy Berg is doing some interesting things with analyzing the data that everyone else has ignored until now.

I find it very amusing that apparently no one bothered to do these kinds of analyses before, or if they did, they weren't allowed to release it publicly?

Now I want to see what Jeremy Berg et al. are going to do about it.

Are there going to be real changes, or is this just an academic exercise where we're still afraid of pissing off the richest people (and yes, that's also a reference to the Obama tax cut nonsense going on right now).

Anyway, we were saying how maybe the real problem is the same old shit about bootstrapping. You can't start a project with no money. Or, you can, but you might have a hard time finishing it and getting it published.

I've always argued that our current funding system places way too much emphasis on reputation and doesn't pay enough attention to who is asking the right questions (especially considering the rampant problems with students and postdocs ghostwriting their PIs grants). It's not a level playing field when you won't let Cinderella come to the dance.

It's no secret that I dislike the current system, where universities take overhead money out of our grants and divert it to god only knows where. And I think it's unfair that you have to be faculty in order to apply for money. Your only alternative is to start your own business, in which case you'll need a lot more money because you'll have zero infrastructure.

My friend was suggesting that there should be money for doing pilot experiments, which would get around this problem that established PIs can use money from other grants to bootstrap future grant applications.

And the evil ones can effectively block their postdocs from doing the same.

I think that pilot bootstrapping facilities & mini grants could be a great solution. I suggested that getting your PhD should be enough training, and then you could go into a pool of qualified applicants who propose ideas and how to test them. Those proposals would be reviewed double-blind. That would be the first round. Then if your proposal was chosen, you'd be given some limited amount of time and money to generate some critical preliminary results, and a chance to present them to compete for an actual grant.

I dunno, something like that.

Point being that, yes, experience would be evident, and it would still be an advantage, but you couldn't check out of the lab completely and still claim to be *doing research*, like some senior PIs. And it would give an opportunity for the creative young folks to get into the game without having to jump through decades of bureaucratic nonsense and political dues-paying.

Still not sure what to do about the university problem, though. It just seems like the current granting system consists of throwing a lot of money into a black hole. And it sounds as though, at least according to Jeremy Berg's data, the more money you throw, the less return you get.

Labels: , , , , ,


At 7:45 PM, Anonymous Anonymous said...

Some fields that NSF funds are very different from NIH in the sense that they are mainly theoretical or based on simulations. Computer science and math are prime examples. Even if the budget is doubled, I don't see them having the same problems as NIH, predominantly because in these kinds of fields, throwing more money at a problem does not necessarily mean more productivity. You cannot prove a theorem faster if you have five students working on it instead of two.

However, what _may_ happen as a side-effect of increased NSF funding is that superstar PIs may start writing many more papers as a result of their increased number of students. Which is perhaps not too bad, and IMO does not offset the good part of doubling NSF funding -- namely, increased funding for more PIs at smaller schools.

At 8:16 PM, Blogger Kea said...

Well, whatever happens, it won't help the old no hopers like us. The current grant I am applying for organises its budget so that postdoc PIs (ie. me) are paid around 1/3 of a full salary. If you try to pay yourself more, the 112% overheads eat up all your capital, and there is none left for conference travel etc. They must think that these PIs will spend 100% of their time doing research, and another 120% in teaching and academic commitee participation.

Of course, I have yet to succeed in getting my own grant.

At 10:06 PM, Anonymous Anonymous said...

You make a lot of points here. Just to address one of them, I disagree with you on grant overhead. Universities get some money from donors, money from the government, grant overhead, and tuition from undergraduates and graduates (paid out of the grants, usually). If grants didn't have overhead, then there would be more pressure to train more graduate students for the tuition. This would not work out well for the "entire generations of trainees not being able to get jobs fifteen years from now." I don't know exactly where all the money is going either, but it does have to come from somewhere, and grant overhead is not the worst possibility.

At 10:15 PM, Anonymous Anonymous said...

I also disagree with you on grant awards. A problem with the current system is that PIs spend their lives writing grants instead of doing actual research. If past results were emphasized more (yes, reputation), then these scientists would be much more efficient.

You will probably reply that this blocks creative young folks from getting into the game. However, I think that there are not nearly enough barriers right now for young people to get into science, and the result is that we have twenty times as many PhDs and postdocs wanting to do research than there are jobs for. It would be better for everyone if there were *more* barriers. For example, a not-atypical department might hire two dozen postdocs every year with on average one faculty opening. This is insane.

There is no easy solution. Adding barriers also hurts, although I think it hurts less than the alternative if the barriers are put in place early. Hitting the wall after one's PhD is painful, but is better than hitting it after four years of postdocs.

At 1:51 PM, Anonymous Anonymous said...

Thanks for your blog, it made me realise I'm not alone in my anger at the way some things work in science! It also made me want to persevere in my blog. You write really well and congrats for keeping it up for so long. I've added you to my blogroll: (another PhD blog!)

At 7:21 AM, Anonymous Anonymous said...

Have you seen the new issue of Nature? Lots of articles about the science PhD and the problematic pipeline to the endless postdoc and no real job.


Post a Comment

Links to this post:

Create a Link

<< Home