Constraint-Loss FMEA¶
Risk analysis — instantiates Latent Constraint Preservation Audit
Enumerates the failure modes that removing a structure would unlock and scores each by severity, occurrence, and detectability to size the loss before the cut.
Constraint-Loss FMEA adapts Failure Mode and Effects Analysis to a single question: if this structure disappears, what can now go wrong? Where ordinary FMEA asks how a component might fail, this variant asks how the absence of a structure lets the surrounding system fail — each constraint the structure was silently holding becomes a candidate failure mode once the constraint is released. Its defining move is scoring the downside: it does not design a replacement or argue for retention, it sizes the loss so a decision can be made on evidence. The output is a ranked table of "here is what breaks, how badly, how often, and how invisibly" — the loss model the rest of the audit prices its choices against.
Example¶
A food-processing plant is considering removing a metal-detector reject station near the end of a cereal line. It rarely fires, it occasionally halts the line on false positives, and the team suspects it is redundant with an upstream magnet. Rather than argue from irritation, they run a constraint-loss FMEA. Each protective function the station encodes becomes a row: a stray bolt from worn machinery reaching a consumer, a metal shard from a cracked sieve, a supplier contaminant the magnet cannot catch because it is non-ferrous stainless. For each, they score severity (a metal fragment in a child's cereal is catastrophic), occurrence (how often such contamination actually reaches this point), and detection (whether anything downstream would catch it if the station were gone — and for the stainless case, nothing would). The rows multiply into risk scores, and the highest — non-ferrous contamination with no downstream catch — is precisely the failure the "redundant with the magnet" argument missed. The FMEA does not decide; it hands the decision a sized, ranked picture of exactly what the removal would expose.
How it works¶
- Turn constraints into failure modes. For each function the structure may encode, state the failure that becomes possible once it is gone — not the structure failing, but the world failing in the structure's absence.
- Score three axes. Rate each mode for severity (how bad the harm), occurrence (how likely the triggering condition), and detection (how likely the loss is caught before it matters if the structure is removed).
- Rank by combined risk. Multiply or matrix the axes into a priority score so attention flows to the modes that are severe, plausible, and invisible — the archetype's most dangerous quadrant.
- Flag the undetectable. Modes that score high specifically because nothing else would notice them are marked as requiring a substitute control or a monitor, and handed downstream rather than resolved here.
Tuning parameters¶
- Scoring scale — coarse (high/medium/low) versus a fine numeric scale. Fine scales rank precisely but invite false precision over guesses; coarse scales are honest but blur close calls.
- Detection weight — how heavily undetectability inflates a mode's priority. Weighting detection hard surfaces exactly the silent losses this archetype fears, at the cost of ranking dramatic-but-visible risks lower.
- Mode-elicitation breadth — how far the team reaches for failure modes: documented incidents only, or imagined edge cases too. Broader reaches catch novel losses but generate noise and can paralyze a low-stakes review.
- Reversibility overlay — whether each mode is additionally tagged by how hard the loss is to undo. Adding it steers toward guarding irreversible losses first; omitting it treats a recoverable outage like a permanent one.
When it helps, and when it misleads¶
Its strength is that it forces the decisive question the archetype is built around — not "what burden disappears?" but "what failure becomes possible?" — and it ranks those failures so effort concentrates on the severe-and-silent ones. The scoring vocabulary is borrowed straight from established practice, where the combined score is often called a Risk Priority Number.[n1] That shared language makes the output legible to safety and quality reviewers who already think this way.
Its failure mode is the unimagined mode: FMEA can only score the failures someone listed, and the loss that sinks a removal is frequently the one nobody thought to write down, so a fully-scored sheet radiates a false completeness. It is also vulnerable to optimistic occurrence and detection scores — the same wishful thinking that motivated the removal quietly deflates the numbers, dropping the risky mode below the action line. The classic misuse is letting the FMEA slide into design — the team, alarmed by a high score, starts specifying the replacement control inside the worksheet and mistakes a proposed fix for a proven one. The guarding discipline is to keep this mechanism strictly diagnostic: it sizes and ranks losses and recommends that certain functions get substitutes, but the substitute is designed, and its adequacy proven, elsewhere. Sourcing the mode list from the dependency and interview mechanisms, not from the analyst's memory, is the main defense against the unimagined-mode trap.
How it implements the components¶
removal_risk_and_loss_model— its primary output: the ranked, three-axis model of what could be lost, expressed as severity, occurrence, detectability, and (optionally) reversibility.constraint_encoding_map— the worksheet's rows enumerate the constraints the structure encodes, one per candidate failure mode, giving a map of what the structure was holding.
It does NOT implement functional_substitute_or_compensating_control — that is its nearest twin Compensating Control Matrix's job: the FMEA sizes the loss and flags which functions need protecting, while the matrix designs the replacement that protects them. Where both touch the shared constraint encoding map, the FMEA reads the map to rate failure, whereas the matrix reads it to assign controls.
Related¶
- Instantiates: Latent Constraint Preservation Audit — the FMEA is the archetype's loss-modeling step, answering "what failure becomes possible?"
- Consumes: Legacy Function Interview and Dependency-Tracing Workshop supply the function and dependency list the failure modes are built from.
- Sibling mechanisms: Compensating Control Matrix · Chesterton's Fence Review Gate · Dependency-Tracing Workshop · Deprecation with Rollback Window · Historical Rationale Reconstruction · Legacy Function Interview · Post-Removal Sentinel Dashboard · Removal Sandbox Trial · Silent Dependency Survey
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Enumerates the failure modes that removing a structure would unlock and scores each by severity, occurrence, and detectability to size the loss before the cut, making its operative form a computation, comparison, model, or analytic representation used to infer, estimate, or choose.
Independent corroboration: The frozen evidence defines Constraint-Loss FMEA as 'Enumerates the failure modes that removing a structure would unlock and scores each by severity, occurrence, and detectability to size the loss before the cut', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Engineering & Design
Origin pattern: Single lineage
Present-day reach: Specialized
Rationale: Reliability and quality engineering cohered FMEA's enumeration and severity-occurrence-detectability scoring of prospective failure modes.
Related originating lineages:
- Operations Research — Risk prioritization supplies quantitative ranking of losses released by removing a constraint.
Review resolution: The mechanism is a direct FMEA adaptation from reliability engineering; quantitative prioritization is a legitimate operations-research contribution, but the underlying method is already recognizable.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] In classic Failure Mode and Effects Analysis, severity, occurrence, and detection ratings are combined (traditionally multiplied) into a Risk Priority Number that ranks which failure modes deserve action first. This mechanism reuses that scoring structure but reframes the failure as the removal of a constraint rather than the malfunction of a part. ↩