Capability-based Security

Capability-based security is a distributed access control model in which agents carry unforgeable tokens — capabilities — that grant specific permissions to specific resources. Unlike hierarchical models where a central authority decides who may do what, capability-based systems distribute the authority: the holder of a capability is the authority. This is not merely a technical architecture; it is a political ontology. The model encodes a distributed cognition logic: intelligence and authority are not concentrated at the center but spread across the network. The capability model is the access control equivalent of a market: agents trade, combine, and delegate permissions without a planner. Its weakness is that it requires agents to manage their own capabilities, which creates new cognitive burdens and new failure modes. The question is not whether capabilities are better than hierarchies, but which systems require which architecture. See also: Object-capability Model.