Frankly none of the five examples strikes me as something I could have predicted, nor ever struck me as such. Nevertheless, social science may indeed produce few significant results which are not predictable. How is that possible given the examples above? Simple: the examples may have been cherry picked to make the point. In particular, their significance (to us now) is seriously damaged by the fact that they are not general statements but are statements about a time and place. While they may be generalizable to the present day while preserving their truth, they may not be. We just do not know. So, as they stand, they are not that useful to us now.

Social science almost certainly produces many insignificant results which are not predictable. It is easy enough to come up with questions which we can then methodically answer by gathering data. What percentage of Massachusetts residents like Fig Newtons? Is it higher or lower than the percentage of New York residents who like Fig Newtons? This is certainly a question that can be asked, and one whose answer I do not know. I could obtain a grant and then spend the grant money studying this question. But it is not a significant question, and learning its answer does not advance human knowledge in a significant way.

But what about significant results? Here’s a much more important question: in general, does extreme lack of sleep tend to have any significant negative impact? This is important because if it does not tend to have any significant negative impact, then many people will find this highly useful knowledge. Many people will sleep much less.

But notice something: it is not only an important question, it is also a question which people know the answer to. And this is no coincidence. It is often hard to hide important truths about people, from people.

This is not true of all social facts. Economic facts are facts about large numbers of people interacting, sometimes very indirectly, and so are a kind of fact which people have a hard time seeing, since they only encounter small numbers of other people at any given time, so they do not see the whole.