Conjecture

Part I states the limit of swarm coordination itself, and it is the same limit our own thesis runs into. If ants had to solve an entirely new survival problem, building a canal or a boat, they could not do it, because novel problems require the intelligence to dream up different approaches and divergent solutions if they are to be solved quickly, and that is what colonies lack. Swarm logic is the product of an environment stable for millions of years, in which trial and error could tune the communication rules. So emergence is excellent at recurring problems and helpless at unprecedented ones. That is the same objection as the fossil record being silent about an inflection point, and it applies to agent swarms and decision markets exactly where we most want to deploy them.

Coordination architecture