Last week I looked at Michael Nielsen’s new book, Reinventing Discovery, and, in particular, how science can deal with the broader citizenry using online tools. As it turns out, that is not where the book begins. Instead, Nielsen starts by looking at how scientists deal with each other.
If you go back far enough, there are plenty of instances where scientists desperately wanted recognition but were reluctant to divulge what for. Galileo, upon discovering what turned out to be rings around Saturn, decided to inform his contemporaries via an anagram (the solution of which would be revealed later). The idea was to buy some time to build on a discovery without risk of being trumped. Newton famously put calculus into a draw before competition forced it out a decade and a half later.
And this continues today. Scientists keep their findings — especially interim findings close to their chests — and nervously build on them hoping they can still claim priority. And there is potentially an efficiency to this. Mike Brown recounts in his enjoyable How I Killed Pluto and Why It Had It Coming, his anguish at being trumped in the discovery of planetoids beyond Pluto while he tried to refine his observations. Those refinements turned out to yield critical pieces of knowledge.
Michael Nielsen wants interim results to be released far sooner than scientists currently do. His reasoning is simple: there are plenty of instances where those results are of use to others and, moreover, the next stage of research is more sensibly conducted by those others. In the language of economics, Nielsen wants more outsourcing and less integration in scientific production. And examples abound of this; especially in mathematics. Here is another example today from Chemistry. Moreover, they are facilitated by new online tools that make it far less costly for scientists to share information. Just think about the efficiency of the speed of review of Ed Nelson’s recent claim.
The argument for diversity and broader attention to problems is a strong one. But, as an economist, one has to ask: why isn’t it a compelling one? Why would scientists bias their disclosures in a way that did not tap that diversity?
Two broad reasons come to mind. The first is plain efficiency. An interim result may be a poorly thought out result. Working that out may claim too much attention that would have been more efficiently undertaken by the initial discoverer. Moreover, communication is costly and do we really want to ask scientists to spend time early on doing that rather than more research and refinement. Put simply, there may be an optimal time upon which results are disclosed and we have to take care in thinking about how to ensure results aren’t disclosed too early as much as we worry about them being disclosed too late.
The second has to do with the price. If disclosure isn’t happening early enough, we have to ask ourselves, why not? If none of the efficiency reasons are driving this, then it surely is the case that the reward for early disclosure is too low. This, of course, is surprising because the reward structure of science favours priority in claims which, in turn, compels disclosure as early as possible. Now it could be that the reward from discovering the ‘whole result’ as opposed to just the ‘interim result’ is high. And so the price we are worried about is too high a reward for the increment building on the interim result to the whole one. But the scientist is compensated there too. They could disclose a result and then go on to build on it themselves; thereby claiming the whole reward. Or, alternatively, someone else does that, in which case the scientist and follow-on researcher share in the reward — with the pioneer scientists being given a valuable citation.
Given this, what has to drive a desire to disclose interim results earlier is a concern that prices in the market for citations (specifically) are not being determined in a correct manner. To be sure, given that there is no real market here and the prices I talk about are subjective and non-transparent, one can believe that they are not right. But are they biased in a particular direction? That is harder to say.
So let’s travel back to economic theory. When might a price be too low? First, the price might be low if there is a hold-up problem. In this realm, that would be a situation where a follow-on researcher has an incentive to reduce the claim of the pioneer researcher and hence, increase their share of the joint rewards. But mitigating this are processes designed to give pioneer researchers some power in staking their claims. So it isn’t obvious that they are victims in hold-up.
A similar argument arises if it is hard for others to evaluate the significance of shared claims. This could be a reason why pioneers receive less than their due. But it equally could go the other way.
That said, these are broad and deep puzzles regarding the economics and sociology of science. The best way to present Nielsen’s argument is to note that online tools have dramatically reduced the costs of collaboration and have, at the same time, increased our ability to quantify contributions — possibly to the point of allocating kudos across shared contributions (using citation indexes and Google Scholar). Those changes alone should given scientists pause to think about whether their current disclosure behaviours — rooted in a pre-networked world — should be changed. Nielsen’s book makes that case well.