I agree. Circular reasoning along with self-fulfilling prophecies are definitely an issue. I still like it for introspection and realizing that other people have different "mind structures", whatever they might be.
> it's not a scientific or testable prediction about the way people act in the world.
The problem is that there is probably no scientific or testable prediction about the complexity of humans. Psychology itself seems not very scientific, so people rely on their intuition (as they should) to understand other people. However that leads to the problem of "X also wants Y, because I want Y", which from my experience is just not true at all. Humans have very different interests and goals. So these models creep up to account for that.
