Taleb formalized this as convex tinkering: small errors with large potential gains. You don’t need to be right often — just keep what works and discard the rest.
Ref
In complex systems, tinkering — trial and error with small mistakes and large potential gains — vastly outperforms planning. You don’t need to be right often, just keep what works and discard the rest. — Taleb, Antifragile
→