Innovation-stimulation: will it work? Definitely worth a try!

Francis Collins has for some reason decided that NIH should try really, really this time, to stimulate research 'innovation' by moving at least a bit away from costly, wasteful, excessive but incremental big project grants, to dedicating a goodly chunk of NIH external research funding to individual investigators rather than large groups.  Here is the Nature story about it.

Alpine ibex climbing Cingino Dam in Italy (source: every other web site)

The NIH has been experimenting with funding high-risk, high-reward science with four separate pilot programs, including the Pioneer awards.  According to the Nature piece,

The NIH currently spends less than 5% of its US$30-billion budget on grants for individual researchers, including the annual Pioneer awards, which give seven people an average of $500,000 a year for five years. In contrast, the NIH’s most popular grant, the R01, typically awards researchers $250,000 per year for 3‒5 years, and requires a large amount of preliminary data to support grant applications.

Expanding the program is clearly a move in a good direction.  Big projects have their place, but have become as much a reflexive strategy for self-perpetuation as they are truly justified by their results history (which, by and large, isn't all that good or has reached diminishing returns).

Of course, individual independent investigators are just people, trend-following herd animals like most of us are.  Once the new program is in place, every investigator will flock to the trough.  Most will propose routine, safe projects even if they assert that they're 'innovative'.

Those proposals that really are innovative will involve risk in two main senses.  First, they mainly will involve procedures or strategies that are to the area and/or to the investigator, truly new, unclear, or untried.  Second, if the work is really innovative, most of it won't get completed on time, won't yield much in the way of publications, and -- worse -- won't find anything really new.

But is that outcome really 'worse'?  We think just the opposite!  If not much is invested in a project, not much is lost if it was truly creative but failed.  By contrast much is currently invested in huge projects that are so safe that they hardly generate commensurate returns.  Indeed, the reason for the failure of a really exploratory study may provide more useful knowledge than most 'positive' studies' findings.  And most potentially innovative ideas are, and turn out to deserve to be, busts.  That is why we call the ones that succeed innovative: they can change how we think.

This NIH policy change won't change crush of competition to keep the grants flowing, and will make it hard to see what is really innovative, in the inevitable panicky rush to get one's salary covered and keep the lab operating.  It takes experience, perhaps, but not undue cynicism, to predict that this new policy will be gamed and strategized.  The overpopulation of investigators still need funding (or jobs!), and will flood to the new trough, finding all sorts of reasons why their work is innovative.   Do you think it could be otherwise, or that such discussions are not taking place already at brown-bag lunches in departments across the country?

Place limits
Unless we limit how much funding any one investigator can have, don't give these new grants to people who already have a grant or impose some such restrictions, we will largely see just be a game of musical chairs.  New labels, same stuff.  After all, who will be reviewing and administering these applications?  It will be the same people who have brought you big-scale non-innovation all these years.  Unless today's heavy hitters are able to reverse the politics back to the old way (making sure their big-projects don't get curtailed!), NIH will make it a new System, with all the bureaucratic politics and cumbersomeness that that involves.  If their career has been spent in the current treadmill, how many will even be able to think in truly innovative ways?  We are, after all, middle-class people who need to earn a living as things now stand. What else can you expect?

Still, the change should be better than what we currently have!  The amount of funds wasted in the new way will be less than the amount being thrown away in the current rush to Big Science, the seeking of huge projects too big to kill and thus to provide career safety for the lucky investigators and their labs and fancy equipment.  As long as in the new way, the funds for individual researchers are enough to let them do good work but not enough to let them get comfortably entrenched, or for their administrators to depend on the overhead, then it's got a chance to make a real difference.

Of course, this will work even better if those who are training graduate students and post-docs inculcate innovative thinking.  If the grants are big enough just to enable faculty to hire students and technical staff to do the work, they may work less well.  What we need are grants to individuals that are small enough that the recipients will actually have to roll up their sleeves and do some of their own work.

We would suggest a fillip that should be tried:  Give grants to graduate students to do their own truly independent project, not just to be serfs on their mentors' project. Independent, free-standing funding for dissertations.  Labs should be their professors' places of training, not just their playgrounds.

Finally, we know that too-few and too-small will not really work well (compare western science to most of Eastern European, Indian, or South American science in the '80s, for example).  But unrelenting vigilance will be required to prevent coalescence once again into fewer, bigger projects.

If it can be done and really done properly, this could be a salubrious change, in directions we have to approve of, since we've been criticizing the current big-science mode for years. But we have to be patient, because innovation is very hard to come by.

Bright people and a coffee pot?

We've just finished a couple of posts on the nature or state of the kind of science that we're largely up to in the life and evolutionary sciences.  We gave a long list of reasons why our precision and rigor are simply far, far below that of physicists and chemists, who have long sneered at the imprecision of evolution and social science.

True, much of biology is not the same kind of science, but there are areas in which social, life, and evolutionary science are every bit as rigorous.  In the social sciences, some aspects of statistical sampling--as in birth and death rates and the like, are very sophisticated.  And of course in many different areas, the life sciences are just as rigorously molecular as chemistry and physics.  DNA does exist, after all, and does code for protein, and sequences do evolve phylogenetically, and so on.

But not just negative!
We complain about the business as usual, or even hyperbolic aspects of those areas of social, behavioral, evolutionary, and 'omic' sciences that are not up to snuff.  But this is not just being negative!  That's because what we know, and what it does not enable us to predict or explain precisely is real knowledge about the world.  That is positive knowledge, even if it's not what we'd like to have found.

But where what's afoot is interpretation and action based on 'judgment', as one of our Commenters suggested (correctly), with poor predictive power, and where our idea of prediction is really based on retrodiction (fitting causal ideas to existing data) and this is not yielding clearly powerful prediction, then this is not like physics and chemistry at the level of daily practice.  Those fields have their blurry edges, but more of a rigorous component.

There is no reason to think the same approaches we've been doing will solve the problem, if the complexity being discovered is real and being interpreted correctly, and the problem is our inability to make precisely specific predictions.  It's here where we question the investment in just keeping the system running, the belief people have in policy makers advised by the same professoriate, and the run o' the mill self-perpetuating university system.  We are not making these issues up!  We just harp on them, and we do that knowing few if any will listen, but perhaps of those who do, somebody will come up with a better way to do things.  You can go down the street and order up a cinnamon latte at Starbucks, but it's not possible to order up genius and transformative discovery just by putting out a request for proposals (RFP) for brilliant novelty! 

There is an historical model for how to stimulate it, however, and perhaps we could try it.

Bell Labs and Xerox
In the mid-20th century a few companies set up basic science labs. The companies had brilliant new products and plenty of ready money (or they were a monopoly and didn't have to worry about the profit bottom line).  The classic example was perhaps Bell Labs, but people also mention Xerox, Land (Polaroid) and a few others in the US, for being places that did manage to stimulate truly innovative thinking.

Basically, some bright (even wacky) people were put in a building where they couldn't avoid bumping into each other on a regular basis--e.g., they had access to only one bathroom. They were given a coffee pot.  And the door was shut.  "Just think about things," was their mandate.  From time to time the boss went in and asked if they'd been actually working, and if so had they discovered anything....even anything useful?   Of course the rest, as they say, was history.  Innovation, not just incremental changes, poured forth.

There are institutes for 'advanced' study (modestly self-named!) in various places.  Most famous is Princeton's because brilliant mathematicians and physicists including Einstein were there (but did real innovation arise, or were people there mainly after their guns had largely been fired?), and there's one at Stanford for the behavioral sciences.  Complexity became a watchword of our time, and the Santa Fe Institute, in which I have an external faculty appointment, were established with initially private funding to tackle the problem of complexity.  There's a new institute called the Evolution Institute  that we've just learned of and been in discussion with, that has innovative ambitions and internet presence, that aims to stimulate interactions designed to use 'evolutionary principles' to do good in the world.  How successful they will be as well as how such principles are decided on or evaluated remains to be seen, but at present they depend on grants and hustling donors, which is a warning sign, because that kind of dependency puts pressure to be predictable and in that sense safe, rather than truly innovative--as university research clearly shows.

Maybe these groups will succeed.  But in our view, success is more likely by attempting Bright People with a Coffee Machine in locked buildings, rather than the basically universal fancy PR, web sites, and the rest of the window dressing, often frankly self-promoting, rather than true, quiet risk-taking.  Endowment with substantial funding and few strings attached is the way at least to try to stimulate such things.  And it should not go to already well-known, award-winning scientists, but somehow work through the middle-class degree mill to find the odd-balls who can really think; many major new industries were founded by college drop-outs--and the distancing from the stultifying atmosphere of universities goes way back and includes some of the most brilliant scientists ever.  They include Einstein, Darwin, Wordsworth, and others.  There's a message there!  But there's absolutely no guarantee.

And if the result of the inhabitants' work is not a preponderance of failures, then they're not trying hard enough, and shut the place down!

Big Science, Stifling Innovation, Mavericks and what to do (if anything) about it

We want to reply to some of the discussion and dissension related to our recent post about the conservative nature of science and the extent to which it stifles real creativity.  Here are some thoughts:

There are various issues afoot here.

In our post, we echoed Josh Nicholson's view that science doesn't encourage innovation, and that the big boy network is still alive and well in science.  We think both are generally true, but this doesn't mean that all innovators are on to something big, nor that the big boy network doesn't ever encourage innovation.  Some mavericks really are off-target (we could all name our 'favorite' examples) and funding should not be wasted on them.  The association of Josh Nicholson with Peter Duesberg apparently has played a role in some of the responses to his BioEssays paper.  That specific case somewhat clouds the broader issues, but it was the broader issues we wanted to discuss, not any particular ax Nicholson might have to grind.

One must acknowledge that the major players in science are, by and large, legitimate and do contribute to furthering knowledge, even if they do (or that's how they) build empires.  There is nonetheless entrenchment by which these investigators have projects or labs that are very difficult to dislodge.  Partly that's because they have a good track record.  But only partly.  They also typically reach diminishing returns, and the politics are such that the resources cannot easily be moved in favor of newer, fresher ideas.

It's also true that even incremental science is a positive contribution, and indeed most science is almost by necessity incremental: you can't stimulate real innovation without a lot of incremental frustration or data to work from.  Scientific revolutions can occur only when there's something to revolt against!

If we suppose that, contrary to fact, science were hyper-democratized such as by anonymizing grant proposers' identities (see this article in last week's Science), the system would be gamed by everyone.  Ways would be found to keep the Big guys well funded.  Hierarchies would quickly be re-established if the new system stifled them.  The same people would, by and large, end up on top.  Partly--but only partly--that's because they are good people.  Partly, as in our democracy itself, they have contacts, means, leverage, and the like.

And it's very likely that if the system were hyper-democratized a huge amount of funding would be distributed among those who would have trouble being funded otherwise.  Since most of us are average, or even mediocre, most of the time, this would be a large expenditure if it were really implemented relative to a major fraction of total resources,  contributions likely watered down even further than is the case now.  But that kind of  broad democratization is inconceivably unlikely.  More likely we'd have a tokenism pot, with the rest for the current system.

Historically, it seems likely that most really creative mavericks, the ones whom our post was in a sense defending, often or perhaps typically don't play in the stodgy university system anyway.  They drop out and work elsewhere, such as in the start-up business world.  To the extent that's true, a redistribution system would mainly fund the hum-drum.  Of course, maybe the budgets should just be cut, encouraging more of science to be done privately.  Of course, as we say often on MT, there are some fields (we won't name them again here!) whose real scientific contributions are very much less than other fields, because, for instance, they can't really predict anything accurately, one of the core attributes of science.

One can argue about where public policy should invest--how much safe but incremental vs risky and likely to fail but with occasional Bingos!

It is clear from the history of science that the Big guys largely control the agenda and perhaps sometimes for the good, but often for the perpetuation of their views (and resources).  This is natural for them to do, but we know very well that our 'Expert' system for policy is in general not a very good one, and we keep paying for go-nowhere research.

Perhaps the anthropological reality is that no feasible change can make much difference.  Utopian dreams are rarely realized.  Maybe serendipitous creativity just has to happen when it happens.  Maybe funding policy can't make it more likely.  Such revolutionary insights are unusual (and become romanticized) because they're so rare and difficult.

The kind of conservative hierarchy and tribal behavior are really just a part of human culture more broadly.  Still, we feel that the system has to be pushed to correct its waste and conservatism so it doesn't become even more entrenched.  Clearly new investigators are going to be in a pinch--in part because the current system almost forces us to create the proverbial 'excess labor pool', because the system makes us need grad students and post-docs to do our work for us (so we can use our time to write grants), whether or not there will be jobs for them.

Again, there is no easy way to discriminate between cranks, mavericks who are just plain wrong, those of us who romanticize our own deep innovative creativity or play the Genius role, and mediocre talent that really has no legitimate claim to limited resources.  The real geniuses are few and far between.

A partial fix might be for academic jobs to come with research resources as long as research was part of the conditions for tenure or employment.  Much would be wasted on wheel-spinning or trivia, and careerism, of course.  But it could at least potentiate the Bell Labs phenomenon, increasing the chance of discovery.

We cannot expect the well-established scientists generally to agree with these ideas unless they are very senior (as we are) and no longer worried about funding....or are just willing to try to tweak the system to make it better.  When it's just sour grapes, perhaps it is less persuasive.  But sometimes sour grapes are justified, and we should listen!

Darwin and Malthus, evolution in good times and bad

Yesterday, again in the NY Times, Nicholas Kristof reported on studies of the genetic basis of IQ. This has long been a sensitive subject because, of course, the measurers come from the upper social strata, and they design the tests to measure things they feel are important (e.g, in defining what IQ even is). It's a middle-class way of ranking middle-class people who are competing for middle-class resources. Naturally, the lower social strata do worse. Whether or not that class-based aspect is societally OK is a separate question and a matter of one's social politics: Kristof says no, and so do we, but clearly not everyone has that view and there are still those who want to attribute more mental agility to some races than to others.

By most if not all measurements, IQ (and 'intelligence', whatever it is, like almost anything else) has substantial 'heritability'. What that means is that the IQ scores of close relatives are more similar than the scores of random pairs of individuals. As everyone who talks about heritability knows (or should know), it's a measure of the relative contribution of shared ancestry to score similarities. It is relative to the contribution of other factors that are referred to as 'environmental.'

Kristof's column points out that IQ is suppressed in deprived circumstances of lower socioeconomic strata. And so is the heritability--the apparent relative contribution of genes. That makes sense, because no matter what your 'genetic' IQ, if you have no books in your house etc., your abilities have little chance of becoming realized. How you do relative to others who are deprived is largely a matter of chance, hardly correlated with your genotype. Correspondingly, there is evidence that scores and heritability rise when conditions, including educational investment, are improved. The point is that when conditions are bad, everyone suffers. When conditions are good, all can gain, and there are opportunities for inherited talent to shine.

Can we relate this to one of the pillars of evolutionary biology? That is the idea, due to both Darwin and Wallace, that natural selection works because in times of overpopulation (which they argued, following Thomas Malthus, was basically all times), those with advantageous genotypes would proliferate at the relative expense of others in the population. That fits the dogma that evolution is an endless brutal struggle for survival, often caricatured as 'survival of the fittest'.

Such an idea is certainly possible in principle. But, hard times might actually be less likely to support innovation evolutionarily. When there is a food shortage, it could be that everyone suffers more comparably, so that even what would otherwise be 'better' genotypes simply struggle along or don't make it at all. By contrast, good times might be good for all on average, but might provide the wiggle room for superior phenotypes, and their underlying genotypes, to excel.

This is not at all strange or out in left field. Natural selection can only select among genetically based fitness differences. If hard times mean there is little correlation between genotype and phenotype, selective differences have little if any evolutionary effect, and survival is mostly luck.

In this sense, from Malthus to the present, this central tenet of evolutionary theory may have been wrong, or at least inaccurate. Adaptive evolution may actually have occurred most in plentiful times, not under severe overpopulation. In most environments, competition is clearly not all that severe--if it were, most organisms would be gaunt and on the very thin edge of survival, which manifestly is not generally true.

The IQ story only reflects some here-and-now findings, not evolutionary ones per se. But it may suggest reasons to think about similar issues more broadly.