I think you should try to get the intent instead of stumbling on surface level. The core idea is that recursion explains emergence.

A distributed system can still achieve centralized outcomes as a result of centralizing constraints acting on it. For example, matter under gravity forces leads to celestial bodies, particles under EM forces lead to stable chemical molecules, genes and species under the constraint of replication lead to evolution, language under constraint of usage leads to the evolution of culture, and brains under the constraint of serial action lead to centralized semantics and behavior. In neural nets we have the loss function as a centralizing constraint, moving weights towards achieving a certain functional outcome.

Ok, so what is the relation between centralizing constraints and recursion? Recursion is how distributed activity generates constraints. Every action becomes a future constraint. I think this approach shows great promise. We can link recursive incompressibility and undecidability to explanatory gaps. You can't know a recursive process unless you walk the full path of recursion, you have to be it to know it. There is no shorter description of a recursive process than its full history.

So what looks like constraints when seen top-down, looks like search seen bottom-up. Particles search for minimal energy, genes for survival, markets search for profit, and our actions for goal maximization. Search acts on all levels, but since constraints are emergent, search is also open-ended.