PM-FLT-0050

A Component model composes the behaviour of the Object models it imports by extending their Filters — binding proposition terms that the Object models left open, so that a State in one imported model gates a transition in another.

Whether that extension may override an imported Filter, or may only refine it, is unresolved and MUST be treated as unresolved.

Non-normative. The blog says "the individual behaviors of the object models are replaced by the behavior required by the Entry Management Component" — which reads as override. Claim 12 says a filter qualifies "a recursive aggregation of quasi-orthogonal object state machines into components of wider scope" — which reads more like composition than replacement.

The difference is not academic. Under override, an Object model's own verification tells you nothing about how it behaves once imported, compositional reuse is lost, and honest impact analysis becomes close to impossible because a parent can silently invalidate a child's guarantees. Under refinement — a parent may only strengthen a guard, never weaken a guarantee — each Object model stays independently verifiable and the composition is checkable. We would prefer refinement, but preferring an answer is not the same as having one, so this stays CONFLICT.

Withdrawn 2026-09-08. The patent description settles it, and settles it the way we preferred — which is the reading we should have been most suspicious of, and is why the quote is now carried rather than the argument. See PM-FLT-0070.