{"id":"navigation-as-language","title":"Navigation as an action language","regions":["computation","representation","shared-inquiry"],"question":"Can navigating an interface become a compositional language for an agent to perform actions?","formulation":"Layered locations and available controls are proposed as a low-token bridge from a general model to specialized actions. A dynamic interface is also considered, so the essential issue is the semantics of movement, not a fixed address scheme.","representations":[{"name":"Location and available transitions","affords":"Makes possible next actions visible in the current state.","loses":"State hidden outside the location can break the meaning of a route.","epistemic_kind":"reconstruction"},{"name":"Explicit action expressions","affords":"Makes preconditions and composition inspectable.","loses":"Adds descriptive overhead and requires a shared interpretation.","epistemic_kind":"reconstruction"}],"tension":"A short route can be operationally useful while concealing assumptions needed for composition.","challenge":"When do two routes denote the same operation, and when does their previous context make them different?","development":"active","epistemic_kind":"reconstruction","evidence":["basis-navigation-as-language"],"next_move":{"text":"Build a small reversible action domain; compare route length, invalid transitions, and successful reuse of an action in another context.","epistemic_kind":"model_proposed"},"open_questions":["Can navigating an interface become a compositional language for an agent to perform actions?","When do two routes denote the same operation, and when does their previous context make them different?"],"formal_status":"open_problem","version":1,"content_hash":"6ca253084aeb705640da3c9d2d5f1cf7e3919d814d04a7297314148fd171c9c0"}