01
Curiosity becomes executable work
On the science surface, curiosity identifies unresolved structure. On the system surface, the scheduler turns that structure into bounded capability requests.
contradiction → formulate discrimination task → resolve required capabilities → find actors and authority → issue lease-bound operations → collect evidence → update question lineage → retire or generate next task
02
Scheduling criteria
- Discrimination valueHow strongly could the result separate live alternatives?
- Capability readinessCan the decisive operation actually be executed and verified now?
- ReversibilityCan the intervention be bounded and undone?
- Evidence qualityWill the result carry enough provenance to update the claim graph?
- Compression debtDoes the current explanation survive only by hiding repair cost?
- Resurrection valueDid an old program become executable because a prior constraint expired?
03
No curiosity without authority
The scheduler may propose work. It may not silently acquire credentials, widen filesystem scope, share a browser profile or convert an observation lease into actuation.
Curiosity chooses where to look. Authority decides what may be touched.