Entropy vs. negative selection
People often blame bad outcomes on negative selection: selection for people with particularly sub-optimal traits, or for particularly sub-optimal behaviour. For instance, in a discussion about explanations of a hypothesised decline in EA Forum content quality Thomas Kwa asks whether there might be an anticorrelation between competence and having time to post on the forum. Similarly, people sometimes argue that certain industries—e.g. politics, bureaucracy, or media—attract particularly dishonest or incompetent people, and that that explains poor outcomes in those domains. Ineffective charity is another example: a common thought is that people are punished for giving effectively, which would incentivise ineffective giving.
Sometimes there is of course such negative selection, but my view is that many are too quick to default to this kind of explanation. Instead, I think that simple entropy—absence of a positive selection—is the explanation much of the time. For instance, I don’t think people are much punished for giving effectively—but I don’t think they’re much rewarded for it either. As a result, their giving is likely not much correlated with effectiveness. And since the most effective charities are much more effective than the rest, that means that their giving is much less effective than it could be.
Similarly, I don’t think there’s much of an anticorrelation between competence and tendency to post on the EA Forum. However, I don’t think that there’s a strong positive correlation either, and that absence of that is enough to explain the lower quality. Other EA content—e.g. books and podcast interviews—on average tends to be considerably better than the EA Forum, in my view. I think the reason is that that content is the result of strong positive selection pressures: EA book authors tend to be particularly competent, as do EA podcast interviewees. In the absence of such strong selection pressure, quality drops substantially.
I also don’t think that politicians, bureaucrats, and journalists tend to be negatively selected for (at least not in Western countries, which I know best). Indeed, I actually think there’s some positive selection for them. To the extent that their work isn’t as good as one would have wanted, the reason isn’t negative selection pressure, but rather that the positive selection pressure isn’t strong enough. Leading politicians are no doubt considerably more competent than the average person, but they’re arguably not as competent as the best potential candidates would have been.
It’s been noted that people have a general tendency to see specific causes of phenomena that actually just are the product of noise. The tendency to see negative selection where there is none may be an instance of that. But I also think that people sometimes underestimate the prevalence of positive selection pressures (e.g. selection for competence regarding important jobs), and that they therefore underestimate the effects of their absence. People often go to great lengths to optimise their environment, and we often fail to grasp the full implications of that. Since many domains are heavy-tailed—the best charities are much more effective than the average charity, and the best researchers are arguably much more competent than the average researcher—the presence or absence of positive selection pressures can make a huge difference.