C.26:4.3 - Escalation by evidence or authority demand

Claim-use class	Use	Required basis
Ordinary FPF pattern	QL is not needed.	Use the ordinary FPF pattern plainly.
QL-lite note	Local diagnosis, model note, or worked recognition.	Fill the short card and, for a quality ascription or model claim, its one-line identity/use boundary.
Reusable pattern prose	A pattern, example, or neighboring note will repeat the move.	Add typed state, probe, or export fields, source support, and local anti-cases.
Decision or assurance use	The claim affects boundary, release, audit, evidence, or work decision.	Add rival explanations, evidence-use class, loss notes, and explicit neighboring-pattern selection.
Ontology or physical claim	A physical substrate, new ontology, or empirical superiority is asserted.	This pattern does not support the claim; use a separate physical or empirical support outside this pattern cluster.
For QL claims that carry decision, assurance, ontology, physical-substrate, or empirical-superiority use, compare rival model families before retaining QL as load-bearing. Failure of a simple Bayesian or passive-read model is not yet evidence for QL necessity; it is evidence for trying richer classical, causal, performative, instrument, active-sensing, or representation-abstraction rivals before QL carries the claim:

Rival family	Use first	Keep QL active only when
Classical Bayesian, nonparametric Bayesian, or ordinary probabilistic update	`C.11`, measurement and evidence patterns, and model-expansion patterns	incompatible sample spaces, contextual probability, order-sensitive query structure, or failure of ordinary total-probability composition remains active.
Causal intervention or ordinary world-state change model	`C.28` for causal use; `A.15`, boundary patterns, and evidence patterns for their respective claims	the intervention is also being used as a read, export, comparison, or optimization of the state it changes.
Performative prediction, strategic response, or dashboard-induced behavior	`C.16`, `A.10`, `B.3`, `C.26.1`, and viability/work patterns	instrument-like state update, incompatible probes, or non-faithful state export remains after the ordinary behavior account is written.
POMDP, active sensing, active inference, or experimental design	`A.3`, `C.16`, `U.Dynamics`, and action-cost patterns	the formal claim also involves incompatible probe frames, contextual probability, or state-representation loss.
State abstraction, representation learning, surrogate modeling, sketching, or ordinary compression	`A.6.3.CSC`, `A.6.3.RT`, `A.19`, `F.9`, and ordinary representation patterns	the shortcut depends on contextual, instrument-like, open-information-system update/probe/export-admissibility, or incompatible-probe structure rather than ordinary abstraction engineering.
Causal abstraction or approximate causal abstraction	Use `C.28` first when the shortcut claims to preserve intervention, explanation, manipulation, or cross-scale structure.	contextual probability, incompatible probes, instrument-like update, open-information-system update rule/probe-frame/export-admissibility, or lossy state export remains after the causal-abstraction mapping between source-scale and target-scale states and interventions is stated.

Math reveal sequence:

Mathematical-formality class	Use	Form
M0 - no math	Everyday FPF use.	Plain-language QL-lite note: false passive read, output, admissible use, and stop.
M1 - structural sketch	A reader needs to see why ordinary comparison or export fails.	Diagram or table: probes, frames, carriers, export loss, unsupported comparison.
M2 - small formal construction	Work a conditional consequence or examine a disputed step on a small model.	A finite-state, matrix or instrument-like construction, with the conclusion and assumptions it actually establishes.
M3 - decision-bearing formal model	A decision relies on a model’s adequacy beyond the small conditional construction.	The assumptions, alternatives and validation needed by that decision, with their failure conditions.
M4 - formal assurance or research claim	The requested assurance or research conclusion requires a fuller formal account.	The reconstruction, proof or data comparison needed for the named claim, including its assumptions and limitations.

Most C.26 use should stay at M0 or M1.

Select the evidence-use class from the question the result must answer. QLP-0 and QLP-1 cover recognition and working support. QLP-2 and QLP-3 supply the additional comparison or assurance needed by a receiving decision. The classes identify useful contributions; they do not require collecting them again when an adequate account is available. Mathematical formality and evidence use are separate choices.

Evidence-use class scales by use:

Level	Use	Required content
`QLP-0` recognition	Example, teaching case, or local recognition prompt.	Claim, example, ordinary FPF pattern, QL cue, and local stop.
`QLP-1` local working use	Local architecture discussion, triage, or provisional design reasoning.	`QLP-0` content plus evidence carrier, time window, uncertainty/confidence statement, and stop/reroute condition.
`QLP-2` decision-bearing use	Boundary decision, bridge/export use, viability move, work claim, or representation shortcut changes what the team should do.	`QLP-1` content plus rival explanations, export/loss note when live, minimal admissible output, selected applicable pattern body, admissible use, and non-admissible use.
`QLP-3` assurance use	A named receiving question requires an assurance conclusion about the QL claim, its empirical adequacy or comparative advantage.	Retain the applicable `QLP-2` comparison and use the `A.10` and `B.3` support needed for that conclusion. Add a `C.16` measurement relation or a Bridge/loss account where the conclusion relies on it. Existing adequate support remains usable; state the limits that affect its use.