It's not a science, and my theory is, it can't be, because it attempts to describe a system of agents who are conscious and therefore know about it, which means it cannot have predictive value, and therefore fails Popper's criterion for science (falsifiable predictions which are not falsified by experiment).

Debatable but I think it's at least mostly true - it cannot be science. Economists are a bit like what used to be called "pundits" in India, and "sophists" in ancient Greece.

There can be good reasoning in it, sometimes, but it's all contingent.