Could Netflix Fix Science? – Best in 2022

First thing is this post is not about Netflix actually doing science but it’s about borrowing a strategy Netflix is using to improve the foundation of science. It could completely transform how scientists spend their time and how science is funded and I’ll explain that. But first the problem which I became aware of, once you’re an established researcher you still have to devote much of your time to asking for money and this is more than just discouraging, it’s actively bad for science. I recently discussed this problem with a couple of scientists who applied economic theory to model this situation.

 

Could Netflix Fix Science?

I think we’ve got it working, the problem that we’re addressing in our paper is the fact that we spend increasing amounts of our time writing grants. We are devoting enormous amounts of time, resources, energy, money what not to the process of doling out scarce funds. Most scientific research is funded by grants provided through competitive proposal contests, researchers write extensive documents about how much money they want and what they’ll do with it and then reviewers lscore them on a set of criteria trying to find the best which will be funded. When this grant proposal system really got going in its current form in the 60s and 70s they were funding about half of the proposals. Now, funding rates have drawn down over the past half-century. The fraction of proposals that get funding has dropped dramatically. The funding rate is often less than 10% so you know you’re spending 80 hours writing something that’s got less than a one in ten shot of actually getting funded in ecology. The funding rate is down to four percent which is not as those funding lines have dropped, it is that people have started putting more and more and more of their time into writing these grant proposals for scientists with low funding rates. Meaning they have to spend a lot of their time preparing their proposals because they need to get into that top 10% or so and this means spending around 20 percent of research time not doing research. But instead writing grant proposals the medical schools people spend up to half their time writing grants there are huge amounts of waste associated with the process of identifying what is the most promising science to fund and this has created an arms race where the more time scientists spend writing grants the more time their colleagues have to spend competing with them at my institution.

 

If you’re writing a proposal above you know a certain dollar threshold. The university has writers on staff who will help you write your proposal and that makes sense from the perspective of the university but from the landscape of science you know across the country that’s the arms race phenomenon. Now every institution is doing this because they’re competing with other institutions and yet the value of the science of the funding program is not changing. The university typically takes a heavy cut from every grant that a faculty member gets. We call that overhead and a typical overhead rate would be about 70% which means that if I get a grant for $100,000 the university gets an additional $70,000. On top of that promotions and prestige are also tied to grant funding. So scientists often spend more time on applications than would be strictly rational just to fund their research. I have had plenty of moments in my own career where I’ve thought geez the one thing I’ve learned above all else since taking this faculty job is how to write a proposal. You know I spent more time thinking about how to write a proposal than I’ve had thinking about how to solve some new math problem or do a new type of experiment. So, I think in some sense we’re training top scientists at top universities to write great proposals. 

Google will soon ban call recording apps from play store

Could Netflix Fix Science?
Could Netflix Fix Science?

The big problem with all of this is that none of it actually has to do with doing science. The grant proposal is not the product grant funding bodies one it’s the science that they’re actually after but instead of doing science researchers are spending their time writing proposals. So, it seems like these contests are not the optimal contests. Now, in funding right here low it just all seems so inefficient to me that I wanted to start thinking about whether you could use economic models to figure out is it inefficient and if so how inefficient is it and what could you do about it. What is the alternative that you’re proposing so the idea is that people submit funding proposals as before but instead of funding the top 10% say you take the top 50% and everyone in the top 50% gets a lottery ticket and then one would just draw a lot from among the proposals that are classified as funding worthy to see which ones actually do receive funding. The thinking is this would improve the efficiency of funding by reducing the amount of time scientists spend writing their grant applications if they know that the application only has to be good enough to get into the lottery round. Well they won’t spend a lot more time making it better there’s no advantage to doing that and so they’ll spend less time writing grant proposals and more time doing actual science. You know if we didn’t have all these personal benefits in terms of promotion and your summer salary and all that what you’d be trying to maximize is just how can I do the most good science possible but some scientists would undoubtedly object isn’t it a bad idea not to fund the very best proposals. When I first read about that, that was a crazy idea and I thought gosh that’s really a dumb idea and then when we started to actually look at the mathematical models around it and we became converts because we saw that. This could actually work if balanced science comes out ahead if you just use random chance the savings of thousands of people’s hours spent doing science instead of requesting money far outweighs the slight drop in idea quality. If that can even be measured accurately from the proposals in the first place but people have definitely studied there’s very little agreement between the referees on a grant proposal panel over which are the best proposals of that top pass. 

 

 

Another solution is to take a page out of Netflix’s playbook. Back in 2006 Netflix launched the Netflix prize. Netflix was interested in improving their algorithm for predicting what movies users of Netflix would be interested in seeing. Netflix offered a million dollars to the first team that beats a certain benchmark in recommending movies. If you like, if you’ve already watched these and liked these then no likely they wanted to open this up to the larger research community because these are problems that people work on in a lot of computer science departments and information schools and so on to see if people could come up with even better solutions then. Motivated lots of people coming in from a lot of different angles to give it a shot with a 10% improvement of their algorithm. The prize was awarded in 2009. So, if instead of funding proposals grant bodies funded completed science. Well then, this would shift the incentives away from the grant applications to actually doing good science. So, good work thinking in all kinds of creative ways about how to design a system of doing science. That works better than the system we have now. Science is on one hand, the greatest invention of mankind in my opinion and on the other hand it is still very contingent. It’s based on, you know, our own cultural history. The scale on which we’re doing science or the kinds of questions we are asking and the sorts of data that we have changed a lot and we need to keep thinking about how to adjust the way that we do science to keep up with that.

 

 

So, I think our generation has a potential to be remembered as a generation that really advanced what it is to do science when incentives are aligned the outcome is new knowledge progress and improving the lives of everyone on the planet but the way funding is currently done scientists waste a lot of their time writing applications for money that they’ll never receive. That’s not only a bummer for the scientists, it’s detrimental to science as a whole and to all of us as well. There are better ways to do this, just ask Netflix.

Leave a Reply

Your email address will not be published.