[llamaimperative]

It's not really true that we have "no" predictive power. Everywhere we've ever looked in any system anywhere in the universe, we've found nothing except the reactions that I've described. Even in biological systems.

We use this knowledge every day to predict and produce new drugs (though again, we are not very good at it).

[bubblyworld]

We have found plenty besides those fundamental reactions - sticking to the theme of evolution, take the concept of the gene in natural selection. In that context, genes and their interactions have enormous predictive power. Fundamental interactions between particles do not.

Genes may be made up of such particles but they are their own concept, with their own interactions, which cannot be reduced to that of their constituents. You could argue that in theory you can measure the state of each particle in all of an organism's genes and simulate them according to the laws of particle physics, but that's not actually possible even in principle, for the simple reason that measuring a system to that degree would destroy it (the uncertainty principle).

There are lots of these "causal blankets" in the world, and I think they're a real challenge to the idea that everything is reducible to particle physics.