My new YouTube video about atheism

[FlatAssembler]

Science isn't just about experiments.

Doing controlled experiments is an ideal of science. But when that's not possible, science should rely on predictions and systematic observations. You know, like astronomy and meteorology do. Inferior to doing controlled experiments, but far better than nothing. Astrophysics is a softer science than particle physics, you don't use astrophysics to contradict particle physics. But the difference between astrophysics and social sciences is that social sciences rarely, if ever, make specific predictions. The only specific prediction made by a social science is, as far as I know, the Indo-European laryngeals.

Einstein wasn't doing any experiments when he came up with the idea of relativity.

Yes, but he was doing experiments to test it.

And in fact, there are certain fields in physics that don't generally involve experiments, such as theoretical physics. It's something similar with social science.

Sure, like the string theory, which most physicists despise as pseudoscience.

that's more a difference in degree really

But, quite often, it isn't. Often times, p-values of 5% or so are a sign of p-hacking. In social sciences, there are almost always multiple null-hypotheses to choose from, and, if there are 20 of them, chances are, at least one of them will interpret your result as statistically significant (if the border of statistical significance is chosen to be 5%). Or, that you repeated the experiment multiple times to get a statistically significant result. But, of course, since the phenomena social sciences study are so subtle and hard to reliably measure, you need to put a high border of statistical significance, or else you will miss most real phenomena.

It's a prejudiced stance borne out of ignorance not of facts.

It's borne out of a-priori reasoning. If some field of study, such as astronomy, uses only systematic observation, and the other one, such as particle physics, uses both systematic observation and controlled experiments, the conclusions of the former are fundamentally less certain than the conclusions of the latter. If chemistry assumes physics is right and the theories in chemistry can only be true if the theories in physics are right, then, almost by definition, theories in chemistry are less certain than the theories in physics. Since biology assumes physics and chemistry are correct, and its theories can only be true if they are correct, they are less certain than physics and chemistry are. Since psychology assumes both physics, and chemistry, and biology, that all of them are correct, then, almost by definition, it's less certain than the softest of those sciences. And sociology and economics assume psychology is correct, and add a bunch of their own assumptions only to make things easier to model (that there are no systematic biases making society irrational...).

Plus, economics being useless is not a prejudice borne out of ignorance if you see that economists almost always talk about fairytales of eternal economic growth, and, when a recession happens, they are seeking for some post-hoc explanation and can't agree on it. It's the same reason people believe computer science is useful that they believe economics is useless: they see the fruits.

---

Is medicine then not true science?

Unfortunately, quite often, it isn't. The general rule appears to be that more it has to do with psychology, less scientific medicine becomes. Psychiatry has a horrible track record of convincing itself that very harmful treatments are useful. Freud convinced himself, and many other psychiatrists, that cocaine is a magical cure for almost all mental illness. The destruction of the frontal lobe was a common treatment all the way up to 1970s, and was counter-productive in almost all cases. And it's even now. Nobody knows how antidepressants are supposed to work, and there is not much evidence they work any better than placebo. The same applies, although to a lesser extent, to neurology. Nobody knows how a low-carb low-protein diet is supposed to help people with epilepsy, nor is there much evidence it actually helps. Though, quite surprisingly, the same is true for a lot of cough medication.

If only statisticians were around here to have a word with you.

Well, I did have some statistics classes at the university. Surveys are the last thing you want to rely on, and are often worse than useless.