Maybe, just maybe...

Jun 26 2015 Published by under Careers, Life

Maybe, just maybe, the NIH/NSF systems we have - aside from the underfunding/glut of PI's - are as good as it gets? No one seems capable of coming up with something better.


21 responses so far

  • The total amount of funds for research and the structure of the research workforce are the problems. We need to adjust one or the other, and preferably both.

  • BenK says:

    Nowhere near. We have seen from other systems and environments that it could be done much, much better. The current system is pathological, but bad money, as they say, drives out the good. The measurement systems are broken, the evaluation systems, the support system, the 'career ladder,' the connections to industry, charity, and teaching, and many other aspects. Asking for more money to keep it going is addressing the broken cup by pouring coffee faster. At some rate of pouring, the hole in the cup seems largely irrelevant - and yet, everyone still gets burned.

    • odyssey says:

      Examples of better systems? And note that I'm referring to the NIH/NSF funding systems, not career ladders, teaching etc.

      • BenK says:

        The Wellcome Trust, HHMI, and even NIH have had positive experiences with 'fund the people, not the project.' However, this model strips out intrusive, expensive, and problematic oversight - which can always be defended - and replaces it with trust, which will inevitably be broken at least once. Doling out dimes, each one carefully counted, is a great way to administer grants and when there isn't anything to compare it to for productivity, when the measure of success is rapid punishment, then it rolls on.

        Meanwhile, we've created all sorts of problematic funding mechanisms based on other silly assumptions. We fund too many 'trainees' which makes them relatively cheap even though they may be inefficient at conducting the research and are being given tasks that are inappropriate to their skills and goals - both pursuing funding too young, and performing tasks that skilled technicians would do with much better success because of practical experience.

        To correct the error of our ways; eliminate training grants. If there are jobs to be had, people will acquire the training to get them. Let PIs figure out the most efficient way to conduct research without additional distortions. If the universities have students who are trying to get a good education and paying for the privilege, then those students will choose mentors based on their own goals and can have some independent expectations.
        When a PI is aiming to teach, he can teach; when he has specific research aims, he should choose the best mechanism for the job.

        Besides, cheap labor is often a substitute for technology and other implements better suited.

        The resulting institutions that would rise in response to funding along the persons not projects model may not look exactly like Janelia Farms, Bell Labs, Cold Spring Harbor (back in the day), the Naval Research Labs, the Rockefeller, PARC, the Field Museum, or any of the well-known successes of the past - but I think that may similarities will carry over, including relative stability, some local depth of focus and specialization, and probably good morale.

  • chemstructbio says:

    Perhaps this is just it. When you are in any crisis situation, any plasible option other than the crisis sounds better than the crisis.

  • Put everything that passes a two-thirds triage into a hopper and select randomly?

    Actually, I think there is no good solution.

  • MTomasson says:

    "If there are jobs to be had, people will acquire the training to get them."
    Yikes. Does that include medical training grants? We subsidize the training of doctors too..

    Any system can evolve. Don't feel the system is good enough as it is. Low funding has exposed serious flaws. Here are my more extended thoughts on the Germain proposal..

    • BenK says:

      MT, Ok, I'll bite.

      My Ph.D. is in Ecology and Evolution; so I'm not as intimidated by your rhetoric as someone else might be. I'll put some of my comments here.

      The short cycle proposal system we have puts far too much energy into generating proposals. Mike (the Mad Biologist)'s strategy does manage to relieve some of that futility and waste; he's said it before and it has always had merit, as does the 'people not projects.'

      People not projects actually has several benefits, one of which is the longer time horizons. The idea is to admit that science takes time, that uncertainty about funding levels is destructive. The time horizons need to be opened up to allow for real discovery. All this starting, stopping, changing direction, it wastes energy.

      The second is that projects cannot be evaluated effectively, that the evaluation and proposal process takes excessive time and effort. The 'people not projects' system does not pick winners, per se, it makes winners - and it makes more winners than the other system, which chews people up, grinds them down, wastes their time, and spits most of them out. On the whole, you can take a slightly refined distribution and give them the tools to do the job well, or you can take more people and give them no straw to make the bricks.

      Evolutionary biology has tons of issues that are still being worked out. Your pet project was 'environmental fit.' Lovely. But what if the environment is constantly shifting? Or being actively shaped (niche construction, including negative niche construction - see Ben Kerr)? Maybe the person doesn't find a fit, maybe he makes it fit (or creates a situation in which he is no longer needed... could be good or bad, depending on various outcomes and time scales).

      Or maybe it's not about the person finding the institution, but people creating the institutions? How about major transitions and those interesting intermediate places where the gap between multicellular and levels of selection mush together? Negative frequency dependence, positive frequency dependence, selection on systems of gene flow (or other methods of trait exchange)?

      As for the statement about medical training - I agree, we shouldn't need subsidies for training when the salaries put one in the top 5% of earnings. Artificial scarcity of training opportunities creates huge selection pressures at the MCAT and match stages. All that tension and waste could be relieved; and then salaries would come down. But the AMA regulatory authorities have a stake in preventing that - and resisting physicians' assistants, nurse practitioners, pharmacists, etc. I wouldn't want to shout "Save the status quo!"

      • odyssey says:

        The 'people not projects' system does not pick winners, per se, it makes winners - and it makes more winners than the other system

        Actually the "people not projects" examples you cited above largely chose people who are winning in the existing system...

        So, how would you chose the people?

      • drugmonkey says:

        The 'people not projects' system does not pick winners, per se, it makes winners

        Glad to see you recognize this but the most accurate configuration is

        The 'people not projects' system picks people and makes them winners.

        The devil is still in the picking of the people to make winners out of. The HHMI system example shows pretty clearly how incestuous, narrow and oligarchical this process is. And that is with the opportunity for HHMI outsiders to come to the attention of the insiders via an alternate system- the NIH Extramural system. The minimal diversity the HHMI system enjoys depends intimately on the much larger and more democratic NIH system. Almost like having a huge farm-ball minor league operation to groom talent.

        So if we convert the NIH system to a insider are outsiders to have any chance at all?

    • odyssey says:

      I agree. The current system was not designed to handle the low funding and high number of PI's we're currently experiencing. Point is, what should it evolve into? I have yet to see a suggestion that would be an improvement over the existing system.

      • BenK says:

        An improvement would be to reduce the amount of administrative overhead and funnel that time, effort, and money into productive research. Make 5 year grants, renewable for 5 at a time, repeatedly. Have a continual watch for fraud and abuse ('for cause') and have a relatively high rate of renewal. If this means that a fairly small population of people are supported, that's unfortunate, but at least it should be the highest proportion of scientists to administrators. Scale the basic support for something like technical 6 reports and 1 admin.

        That is much more detail than is necessary for me, but it is at least one concrete alternative.

        • drugmonkey says:

          How do you select your "fairly small population of people"?

          • BenK says:

            I'm not proposing that we have a radically different way of screening them; performance in graduate school, recommendations, perhaps a postdoc for early investigators, or some industry or government research experience; a monograph describing a proposed area of research and some initial plans.

            If you want to give everybody unicorns and rainbows, oh joy! I would at least redirect the existing funds to useful research, rather than administrative overhead and proposal writing/reviewing exercises. I don't think the devil is in picking who will be a winner. The current devil is in the fact that the process of choosing ends up making losers of most everyone; even those that win are typically hobbled or hamstrung.

            I don't see the system working the way you describe it, frankly, with the farm team and so on. If you wanted a farm team, then perhaps you could create such a thing, with low stakes and low overhead. But that isn't what we have now.

            I also don't buy your argument about 'scale.' I think you need to define your terms. When solutions don't 'scale' it is because of some 2nd order effect that gets worse non-linearly. I don't see any intrinsic systemic flaw that causes the 'people not projects' system to crash as you add people and money to it.

  • qaz says:

    The NIH and NSF systems worked fantastically well when they were fully funded. Rather than fighting to redefine the systems, let's get them fully funded. As a function of GDP, we spend less than a third on science than we did in the 1960s.

  • DJMH says:

    I think the means of picking research is as good as it can be. I just wish the NIH would switch to the NSF model of funding, ie total budget not direct budget, so that all institutions would have some reason to keep their IDC rates low-ish. That's all.

  • […] boils down to give money to folks like "me" and fuck everyone else. So I have to agree with Odyssey's premise, maybe the current system is as good as we get. Maybe, the current system just cannot function […]