A.19.ULSAM:3 - Forces (normative)

• Lawfulness vs convenience. The most “convenient” aggregation (e.g., weighted sums) is often illegal across scales/units; lawful folds require explicit constraints.
• Explicitness vs brevity. A single scalar is short to discuss, but expensive in hidden assumptions.
• Kernel stability vs method evolution. Aggregation methods evolve; the kernel must not.
• Evidence gating vs “always return a number.” The mechanism must support abstain/degrade rather than coercion.
• Optional stage vs pipeline clarity. fold_Γ? is optional in CHR protocols; optionality must be explicit (not implicit “sometimes scoring folds”).
• Auditability vs minimal overhead. Recording contributor sets and effective pins adds overhead but prevents semantic drift.
• Declared-set locality vs reuse. A fold is valid for one admitted measure set, grouping or membership basis, policy editions, scope and window, evidence basis, contributors, and intended result; a later use must recheck those premises and cite any relation it actually relies on.
• P2W separation and gate/guard separation. ULSAM must expose eligibility and audit pins without turning into (i) a WorkPlanning baseline binder or (ii) an admissibility gate: planned slot fillings belong to WorkPlanning plan items, while GateDecision/GateLog live in gate patterns / WorkEnactment (suite protocols remain mechanism-steps only).