Ever since Trump’s election, I’ve been reading a few conservative websites.
David Ng

The central problem is that the practical value of “real facts” is often small, compared to the noise. Not necessarily a problem specific to politics either: in many instances, the most important role that a team leader can play in a project is to unambiguously point to a direction and get things moving, not necessarily moving in the “right” direction, because not only nobody knows what the “right” direction is in the first place, a lot of different directions become the “right” direction if people put in the work. (There was a magazine article (or an interview) by Fred Brooks, the guy who wrote The Mythical Man-Month, that made this very point about IT teams. But this applies to almost any organization, I think.) Of course, this is the coalition problem restated in slightly different context: the slogans and symbols (let’s face it — that’s what the “coalitional facts” actually are) are there to get people working together for a common goal, not because they are “true.”

This is, in a sense, a problem that’s bugging me a lot lately: making things work, especially in everyday context, does not require good science, or even any science. The “truth,” for its own sake, at a conceptual level, is pretty much expendable. There is something deeply correct even if disturbing about the bold proclamation, a la Wired magazine, that science is now obsolete since we have so much data. This is not necessarily a different problem from the coalitional problem — in fact, the quantitative measurements of liberalness and conservativeness people use are literally based on this view: we don’t know what a liberal or conservative is, but we know who conservatives and liberals are and how they voted (or responded in surveys), and we estimate how liberal and conservative others whose “ideology” we don’t know by how much they agree with the known “conservatives” and “liberals.” The standard ML approach, although this approach was developed in 1980s and popularized in 1990s, about a decade before ML became big.

The approach itself mimics what the coalitional problem underscores: we don’t know what a conservative/liberal is, but we know who are conservatives/liberals, and the (other) conservatives/liberals are simply those who hang out with the known conservatives/liberals, regardless of what they “really” believe. However unsatisfying it is for those who want to know “what”conservatives/liberals are, it does do a wonderful job of making accurate predictions though, better than any theory/model that tries to address the “what” question, which suggests that the anthropological perspective is exactly right:

With regards the “facts,” I am actually fairly confident that no educated creationist actually believes in “creationism,” at least not a version that qualifies as a scientific theory — that is, something that is falsifiable. There are actually a lot of educated creationists (the average creationist is actually more educated and does better on science factoid quizzes than non-creationists) and they are remarkably good at finding ways to accept both useful and practical science applications (e.g. vaccines, livestock breeding, gene therapy) without actually going back on their creationist beliefs. There is no point in trying to convince them that their beliefs are “false” because it is simply not falsifiable. It’s not a “theory.” It is so flexible that they can accommodate anything that is actually useful and practical. But, on the plus side, this also suggests that, if one treads carefully, this is something that does not actually need to be an impediment to science education. Stephen Jay Gould took a lot of flak for suggesting non-overlapping magistria between “science” and “religion,” but, from the social science and philosophical perspective, Gould seems to have had the right idea after all. Science does not rest on the conclusions: it rests on theorize, observe, re-theorize, re-observe, rinse and repeat, and whatever conclusions there are, they are only transitory, just the best explanations given the observations currently available. Coalitional beliefs are exactly the opposite: it only requires some inviolable symbols that show that those who adhere to them are “good people,” i.e. on the same side. They are not even in the same plane.

In a sense, this is a repetition of the Chinese Rites controversy: the Dominican and Franciscan missionaries in China insisted that Confucianism should be condemned as paganism and the point should be pressed loudly that Confucius is in Hell, while the Jesuits felt that most of Confucianism was not even “religion” in the Western sense and was no hindrance to Catholicism, as long as Chinese Catholics are instructed to adhere to only the “civic” aspects of Confucianism — just enough to show that they are good and respectable members of the Chinese “coalition.” A lot of bruhaha came out of this mess for several centuries. In the end, I think something similar needs to be done, not just so that science could skirt unnecessary controversy, but for the sake of science itself, as per the article I linked to.