When Kestrel County debated how to respond to conflicts between cyclists and pedestrians, two approaches dominated the discussion. The first approach relies on centralized control: a small team sets standards, monitors results, and enforces corrections. Its strength is consistency, but it can miss local conditions because it treats every site as interchangeable. The second approach relies on distributed judgment: local teams test small changes and share what works. It adapts quickly, but outcomes vary unless there is a clear way to compare results. Choosing between the two depends on whether the problem is mostly uniform or highly dependent on context.