It's not nice to fool your colleagues

I don’t think it takes much philosophical sophistication to understand the challenge we face in defending the norms of science against infection by the norms of politics. Mostly you just have to be specific.

Take the report out this week about the paper that was just retracted from Science. According to Retraction Watch the political scientists who did the study claimed to find that “short conversations could change people‚Äôs minds on same-sex marriage.”

1. In practice, there isn’t any important ambiguity associated with the notion that this claim either is true or is not.

2. I also don’t see any subtlety about the idea that when science works well, it makes progress toward a consensus about the truth of such a claim. To be sure, when someone fakes data, as the Ph.D. student on the Science paper did, the consensus might be sent off in the wrong direction. But when science works, there is a tendency toward self-correction that moves things back in the right direction. In this case, the senior author on the paper retracted it after graduate students at another university who tried to replicate the results couldn’t do so, and when challenged, the junior author could not produce records about how the data had been collected.

3. I do not understand anonymous Bitcoin/comment-thread version of science that some people seem to propose? Is there any evidence that this has ever worked?

If there is no notion of individual reputation, people like the student on this paper can flood the world with anonymous papers based on made-up data. The claim that we should just evaluate each piece of evidence and each bit of reasoning its own merits with no regard to who the author is seems to assume that it is costless to detect a fraud, or at least relatively low cost compared to cooking up a fraud. I think that people who make this kind of claim underestimate the prevalence and effectiveness of lies and deceit in science, and how much more prevalent they could become if science becomes more like politics. (See below about what happened to John McCain.)

4. If reputation matters, then there must be some people who have good reputations that scientists listen to and some who do not whom scientists ignore.

5. The equilibrium that I cited, in which one demonstrated instance of fraud or deceit means that peers shun the perpetrator forever after, seems obvious. It is what we see in practice. It is what you’d predict from the standard model of reputation based on agents of different types. I predict that Princeton, which made a job offer to this student, will retract the offer; that the student is unlikely to write any more papers in political science; that if he did, they would be unlikely to be published; and if they were published, they would rarely be cited.

6. To prevent this type of fraud, I do not think that it is feasible or helpful to try to screen people who want to participate in a scientific discussion for “the right” political views.

7. I imagine that the student who committed this fraud has political views about policy toward same-sex-marriage but I don’t have any idea what they are and I don’t see how knowing what they are should change how we respond to a case like this after the fact or what we should do to prevent other cases like this in the future.

8. It is the norms of politics that we should worry about, not political views.

Consider this hypothetical. A Ph.D. student who seeks a job as research assistant for an economics professor explains that he spent several years before graduate school working as for a political consultant who specialized in spreading false rumors. In the interview the candidate speaks with admiration about most some of the most famous anonymous smear campaigns, such as the one in the Republican primary campaign in South Carolina in 2000 that used used push-polls, emails, audience plants, and printed flyers to spread the lie that John McCain had fathered a child out of wedlock with a woman who was black. He brags that he is the Chicago troll on EconJobRumors.

How will the professor react to this information? By being more cautious than usual about trusting this student to do unsupervised data collection? Or instead by saying, “tell me more about what you can do with this EconJobRumors site.”