Yesterday, I talked about why Popper was wrong about how to tell science from pseudoscience. Today I'm going to contradict myself and talk about why he was right, and why I think climate science is still science.

The reason Popper is so tantalizing is that he was clearly on the right track. If your scientific theory cannot be falsified no matter what happens, it doesn't make useful predictions. Astrology, for instance, can never be proved wrong. If an astrologer tells you you'll fall in love and you don't, that doesn't invalidate astrology, it just shows that she forgot to consider that Venus was in Sagittarius or something. There's clearly something fishy and nonscientific about that.

The problem with Popper's account is that he thought that there would be one moment, one experiment that would show that something was right or wrong. But science doesn't work that way -- it takes a lot of little niggling problems to show up that don't quite fit with the theory, and then some new theory will come along that explains all those things. This can take a long, long time -- it took 230 years for Newton's theory of gravity to get replaced. Or it can happen quite quickly, especially for new theories in experimental fields. But in order for those niggling problems to accumulate, there must be testable predictions, things that can be wrong. If your theory is so broad that it never can be proved wrong, it's not science.

This is why I think climate science is doing the right thing, and I'm more skeptical about evolutionary psychology. Climate scientists produce lots of testable predictions: that's what climate models are for. It's true that it's very hard to falsify any particular model, because the predictions are noisy and the actual results are noisy, and so they don't have to be all that close to overlap. In other words, one or two years of results aren't going to be convincing to someone who thinks the model is doing something right.

On the other hand, over a span of several or many years, it will become clear which models are working well and which ones are not. Even now, models are going obsolete every year as predictions get refined, and as models take into account more variables and gain more granularity. I certainly think there's some chance that the current warming trend is temporary variation, though I think it's small. I think there's almost no chance that the field is going away, or that it will somehow fail as a branch of science.

On the other hand, I'm not so sure about evolutionary psychology. I know a lot less about it, so I could easily be wrong, but as far as I know, they mostly use it to explain existing facts. As in, "women seem to be more social and less visio-spacial than men, let's think about our evolutionary history and see if there's an explanation there." The problem with that is that it's very easy to look at a super complex system like our evolutionary history, especially when it's based on ideas of how we lived in prehistory for which we don't have direct evidence, and explain anything you want. My sense is that if we discover tomorrow that girls are better at geometry after all, we could pretty quickly come up with an evolutionary psychology explanation for it.

Of course, there certainly are some testable predictions. The biggest success is in cheater detection -- evolutionary psychologists came up with a novel explanation of a puzzling psychological result, which is that people have a natural propensity for figuring out if someone is cheating on a social contract. Then they tested their theory with new experiments against the existing theories, and were shown to be right [1]. But even this is mostly a new explanation of an existing result, not really a new prediction of human behavior.

On the other hand, evolutionary psychologists are pretty clearly on to something. We know that evolution can change mental abilities, so you would certainly expect that different evolutionary environments would cause different outcomes. It's also pretty clear that our bodies and brains evolved in a very different environment than the one we now occupy, so there's bound to be some conflict, which causes things like high obesity rates due to overeating, or jealousy and anger being too powerful emotions for anyone's good. Evolutionary psychology is trying to explore that area, and there are some very smart people coming up with interesting ideas.

The problem is that those ideas are very hard to verify. History is replete with smart people coming up with interesting ideas that are widely accepted and also wrong. Pick a common idea from the opposite side of the political spectrum from you, and you'll see what I mean. Just because lots of evolutionary psychology sounds good to me, doesn't mean it's right -- and my concern is that there isn't a good way to validate it.

Unlike climate science, where validation (or invalidation), for skeptics, is certainly coming, in just a decade or two.

[1] The test is cool, actually. There are two versions of the test, which are isomorphic. The puzzle is that only about a quarter of people get the first version right; three quarters of people get the second one right. Do it yourself: test one, test two. Take a look and see what you think -- it'll only take a few seconds.

Done? Most people choose cards 1 and 3 for the first test, though looking at card 3 doesn't prove anything, and you really need to check card 4. On the second test, though, people look at cards 1 and 4. The current theory is that we have evolved neural circuitry for detecting cheaters, and we use that for the second puzzle but not the first.