He doesn't get into this, but I think the reason economics is doomed to never have predictive value (as currently practiced) is that the premises are subject to change, and are continually being manipulated by people seeking competitive advantage.

Let's take gravity as an example. Suppose it were not a natural law but a collective agreement we all use to keep each other (and ourselves) anchored to the ground. But we all want to be just a little bit more firmly anchored than anyone else. If we have access to the mechanisms through which gravity is exerted, some people will attempt to anchor themselves at the expense of their neighbors.

Long term, it's better for everyone that we all stick to the ground, since there's the most overall gravity exerted. But there will be local maximums, where some people dig underground as their neighbors float off into space.

As long as small, powerful groups or individuals have the ability to exert control over larger groups, the only valid short-term predictions will concern the behavior of these predatory groups.

The only way to control short-term atavistic behavior is to change the system itself, limiting the scope for individual actors to drive large scale changes.