New Incentives will not Save Science

By Alexey Guzey. Alexey is the Executive Director of New Science.

Published 2021-11-03.

Firms tend to be structured pretty similarly. They all have employees, managers, managers of managers, HR departments, and so on. If we take two firms with identical organizational and incentive structures that work on the same things, some of them will become Stripe and Facebook and some of them will end up on the TV show The Profit (MR).

Similar things can be said about countries. You can copy all of the US’s laws and structures of government and this will absolutely not lead to your country’s GDP per capita suddenly (or ever) jumping to $60k/year.

I noticed a similar pattern of illegibility when writing about the NIH in How Life Sciences Actually Work: looking at the titles of research proposals that get funded and at people who get funded is informative — but only to an extent. You’ll never learn just by looking at NIH’s procedures and statistical data on grant-making, for example, that a significant portion of those grants are written by grad students and postdocs with their PI’s names stamped on them, or that the projects actually pursued with the grant money differ quite a bit from what the grant proposal says on paper, with a don’t-ask-don’t-tell dynamic between the NIH and the scientists, allowing them to work on things the NIH wouldn’t officially fund, while submitting proposals for projects that are almost complete anyway.

Laws, people, and culture are all in a complex relationship with each other, all shaping each other, and just coming up with better incentive structures will not save you, as what is written down is not what really happens, even in societies with lots of written records like ours. Rules are selectively enforced or just not enforced at all. Rules do matter, but only as far as people responsible for enforcing them care, and this, unfortunately, is more rare and more difficult to achieve than it seems.

Even in crypto, where “code is law”, ultimately, when something that stands within the letter of the law but offends the spirit of the law (like finding an exploit in a smart contract), the tendency is to follow the spirit, rather than the letter. And spirit is determined by people partaking in the community and the culture these people create.

In my mind, all of this points to primacy of culture over structure and to primacy of studying the illegible over the legible, leading, in turn, to my reluctance to endorse some grand ‘seeing like a state’ plan for rebuilding scientific institutions. I used to obsess over mechanism design and incentive structures, but I’ve come to believe that we are Bob-Taylor-constrained much more than we are the-ideal-organizational-and-incentive-structure-for-a-basic-research-institution-constrained.

Even though there are clear structural improvements to be made — like letting really talented people without degrees start their own labs — New Science is unlikely to be trying to carry out one single plan, like introducing UBI for scientists, abolishing academic journals, and returning academics to their natural Rousseau’s noble savage-like state or anything of this sort.

Instead, I believe that the most promising way to achieve large-scale improvement in the way basic scientific research is organized is to start small, help individual scientists, and to take small steps towards a much better world.

HHMI Janelia serves as an instructive example to these points. It’s an independent research institute, enabled solely by philanthropic funding and explicitly modeled on Bell Labs and Cambridge’s Laboratory of Molecular Biology, that has made significant contributions to biology since its opening in 2006. Yet, while Janelia has become one of the world’s premier research institutions and serves as an important proof-of-concept of what’s possible to achieve in less than 20 years, it seems to have become, perhaps a bit unusual, but mostly an outgrowth of the established academia, tightly integrated with the rest of the system. This is despite explicitly being designed and structured to be entirely different, something of a Bell Labs of biology.

One part of an explanation for why this happened is that people who work at Janelia are scientists who came from academia and who expect to return to academia after just a few years. This means that, in the end, their accomplishments at Janelia are evaluated not by those trying to do something differently, but by those deeply within established institutions.