Kolmogorov complexity. Chemistry is made of physics; it's necessarily more complex than physics, thus a more difficult topic of study. Biology is made of chemistry, psychology is made of biology, etc.
Why are all the smart guys in physics?
A) nobody likes challenging themselves. Smart guys can see all the easy successes are in the easy sciences.
B) in the easy sciences, it's easy to catch frauds, so by process of elimination all the frauds (dumb) end up in the difficult sciences.
This means scientist quality is inversely correlated with difficulty of subject, and thus science quality is exponentially inversely correlated with subject difficulty. The greater skill required to catch a fraud is matched by a lower average skill in the field.
E.g. in chemistry, scientist quality is lower, meaning theory quality is exponentially lower. However, catching frauds using these simple theories is still sufficiently easy, so there isn't much in the way of frauds, merely poor understanding of the topic.
E.g. in the topics-previously-known-as-philosophy, it is near-impossible to expunge fraud, because a clear demonstration of truth and falsehood requires very high skill indeed. Result: topic is fully dominated by fraud.
Exception: sociology breaks the pattern. In the same way that averaging a bunch of quanta gets you classical physics, averaging a bunch of psyches gets you sociology, which is dead easy. At this level the complexities cancel each other out rather than crystallizing toward further complexity. I suspect there's some evolutionary booster rockets as well - the brain has relevant ASICs. Perhaps even evolutionary sociological grooves, containing sociological responses to a small subset.
Indeed sociology is so easy peasants go, "It can't possibly be that easy," and discount the obvious solution in favour of someone who sounds fancy.
E.g: what if getting taxed makes a country poorer for the benefit of a small, unproductive minority. Mind blown yet?