Γ‑fold is a declared governing spec ref, not an implementation choice. In FPF terms, “how we fold” is a policy-level commitment: GammaFoldSlot MUST be resolvable to CGSpecSlot.Γ_fold routing or an explicit pinned override. If you cannot cite it, you do not have a fold — you have a hidden default.
ULSAM is not normalization. ULSAM does not establish comparability by itself: it does not normalize, rescale, or “align units” as a hidden convenience. If a compare‑on‑invariants surface is required, invoke UNM explicitly upstream and cite the effective pins in Audit.
Prefer vector semantics when possible. If you do not strictly need one aggregated measure, keep measures separate and let CPM + SelectorMechanism operate on a partial order (set-return semantics). A fold can discard distinctions; state which distinctions the declared fold preserves and loses.
Contributor surfaces are not “nice-to-have” in practice.ContributorSetSlot? is optional in the signature, but operationally it is the simplest way to prevent “mystery rollups” and to preserve an explanation surface.
Time-fold is a specialization, not a loophole. The base ULSAM declares Γ_timePolicy and allows time-fold only via explicit windowing policy. If a project needs an explicit FoldTime_Γ operator, introduce it as an ⊑⁺ extension with no mutation of inherited operations or SlotKind drift.
Use the suite lexicon token GammaTimeRuleSlot for the additional windowing rule input; do not overload GammaFoldSlot or invent a generic context input to carry time semantics.