The declared Commonplace frame

Type: kb/types/note.md · Tags: computational-model, self-improving-systems

Self-improvement, reflectivity, and autonomy attributions are frame-indexed: the bearer is a bounded system, not a substrate, so an attribution is elliptical until its boundary is named — the semantics is stated in the self-improving system definition's Provenance. This note is the single citable declaration of Commonplace's own frame. Applications cite it instead of re-deriving the boundary; an analysis that needs a different boundary declares its own frame explicitly and names the departure, rather than silently drifting from this one.

The boundary

Inside: the repository and its operative artifacts; the software and agents that consume them — commands, validators, the review store, harness-loaded agents; and designated maintainers acting in their established improvement roles.

Outside: arbitrary contributors, readers, and advisers; the model provider and its weights; inference infrastructure; hosting.

The partition

Within the frame, the repository, commands, validators, review store, and agents are the computational components; designated maintainers are the human components. The partition reports actor allocation and nothing else — it changes neither membership nor reflectivity.

Using the frame

  • Any membership, reflectivity, closure, or allocation claim about Commonplace is read against this frame unless the claim declares a different one. The assessment horizon is declared per attribution; the frame fixes who and what, not when.
  • Provider models and weights sit outside, so Commonplace has no reflective coverage of their distributed-parametric form under this frame. Requested model bindings are localized configuration inside the boundary: the repository makes the request explicit and revisable, providing a possible selection-control surface over an external dependency when the harness honors it. The six-path audit shows that requested-to-realized binding is not currently trustworthy.
  • Computational-closure readings under this frame count human dependencies only. How a hosted model's dependence on outside inference infrastructure should read is addressed separately in the closure note.

Relevant Notes: