Nice post. I don't know why there is no a bigger push in experiments/investigations with tree-like massive combinatorial expansion with lazy-evaluation, so the

Nice post. I don't know why there is no a bigger push in experiments/investigations with tree-like massive combinatorial expansion with lazy-evaluation, so the solution space can be somewhat "reasonable" (e.g. assuming you have infinite combinatorial expansion capability, adding restrictions, so once you start the evaluation most of the combinations get discarded, so you'll have to deal with a "sparse combinatorial expansion" analog to a convex space for solutions in geometry).