Loss-Channel Decomposition¶
Decomposition analysis — instantiates Realized-Possible Outcome Gap Mapping
Breaks a single measured realized-possible gap into named loss channels that sum back to the whole, so a lump deficit becomes an itemized account of where the outcome leaked.
Loss-Channel Decomposition takes the one number a gap table produces — "we are 16 points / 40% / $2M short of what was possible" — and splits it into a mutually-exclusive, collectively-exhaustive set of channels that each account for a piece of the shortfall and sum back to the total. Its defining commitment is the additive ledger: every unit of gap must be assigned to a named channel (a bottleneck stage, a leak, a mis-specification, a measurement artifact, or an explicit unknown remainder), and the channels must reconcile to the measured whole. It says nothing about how big the total gap is (it inherits that) and nothing about which channel to fix (that comes later); its entire contribution is turning an undifferentiated deficit into an attribution most improvement work never bothers to build.
Example¶
A semiconductor fab reports that a mature product line is running at 71% die yield against a possible ~89%, an 18-point gap. "Fix the yield" is not an actionable instruction. Loss-Channel Decomposition builds the account. Using in-line inspection and end-of-line bin data, the 18 points resolve into channels: 7 points to a specific particle-defect signature at the lithography step, 4 points to edge-die exclusion, 3 points to parametric failures at final test, 2 points to a wafer-handling scratch mode — and a 2-point unknown remainder that no current measurement can attribute, carried explicitly rather than silently folded into a larger channel.
Now the debate is not "why is yield low" but "which of these five channels is worth attacking, and how sure are we of each split." The particle-defect channel, at 7 points and tightly measured, becomes an obvious candidate; the 2-point remainder becomes a flag that instrumentation, not process tuning, is the next need there. The decomposition changed nothing physically, but it made the gap legible.
How it works¶
- Choose a partition that reconciles. Pick channels — by stage, by loss mechanism, by cause type — that are mutually exclusive and jointly exhaust the measured gap, so the parts sum to the whole with no double-counting.
- Attribute with evidence, not narrative. Each channel's size comes from instrumentation, staged measurement, or controlled comparison, not from a plausible story about where the loss "must" be going.
- Keep an honest remainder. Whatever cannot be attributed with current evidence goes into an explicit unattributed channel rather than being smeared across the others to make the picture look complete.
- Band every channel. Each channel size carries its own uncertainty band, so a confidently-measured 7 points and a hand-waved 3 points are not treated as equally solid.
Tuning parameters¶
- Partition axis — decompose by pipeline stage, by loss mechanism, or by cause category. Stage partitions localize where; mechanism partitions localize why.
- Channel count — a few coarse buckets or many fine ones. Finer channels pinpoint but demand more measurement and risk spurious splits.
- Remainder tolerance — how large an unattributed channel you accept before pausing to instrument. A low tolerance forces measurement investment; a high one ships a vaguer account.
- Attribution rigor — from expert allocation to controlled measurement per channel. Higher rigor narrows each channel's band but costs time.
- Reconciliation strictness — how exactly channels must sum to the measured total before the decomposition is trusted.
When it helps, and when it misleads¶
Its strength is that it replaces a single dreaded deficit with an itemized, summable account, letting attention concentrate where the loss actually is instead of spreading evenly across every suspected cause. The explicit remainder is a feature: it marks the boundary of what is currently understood.
Its failure mode is attribution by story — assigning gap to channels because they are familiar or blameworthy rather than because they were measured, which manufactures a tidy waterfall that sums perfectly and is quietly wrong. A classic misuse is collapsing the unknown remainder into a named channel to avoid admitting ignorance, which then sends fixes at a cause that was never really there. The guarding discipline is to treat the decomposition like the loss accounting behind Overall Equipment Effectiveness[n1] — every channel evidenced, the parts forced to reconcile to the measured whole, and the unattributed portion carried openly rather than hidden.
How it implements the components¶
gap_decomposition_scheme— its core output: the mutually-exclusive, collectively-exhaustive partition of the measured gap into named, reconciling loss channels.uncertainty_band_for_gap— each channel carries its own band, and the unknown-remainder channel is where measurement uncertainty is explicitly parked rather than smeared across attributed channels.
It does not measure the total gap it splits — outcome_of_interest_boundary, realized_outcome_record, and realization_gap_measure belong to its nearest twin, Realized-Possible Gap Table, which produces the single number this mechanism decomposes. It also does not judge whether any channel is worth closing (closability_classification — that is Closability Scoring Rubric).
Related¶
- Instantiates: Realized-Possible Outcome Gap Mapping — supplies the attribution layer of the gap ledger.
- Consumes: Realized-Possible Gap Table supplies the single measured gap this mechanism partitions.
- Sibling mechanisms: Realized-Possible Gap Table · Feasible-Frontier Mapping · Best-Demonstrated-Practice Comparator · Counterfactual Ceiling Probe · Closability Scoring Rubric · Gap-Closure Experiment Backlog · Theoretical-Ceiling vs Feasible-Target Review · Post-Closure Gap Remeasurement
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Loss-Channel Decomposition operates as a computation, comparison, model, or analytic representation used to infer, estimate, or choose because it breaks a single measured realized-possible gap into named loss channels that sum back to the whole, so a lump deficit becomes an itemized account of where the outcome leaked.
Independent corroboration: The frozen evidence defines Loss-Channel Decomposition as 'Breaks a single measured realized-possible gap into named loss channels that sum back to the whole, so a lump deficit becomes an itemized account of where the outcome leaked', so its operative form is Analysis, Modeling & Optimization.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Engineering & Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: The mechanism is anchored in manufacturing and industrial engineering, especially OEE-style decomposition of output gaps into loss categories.
Related originating lineages:
- Accounting & Auditing — The additive ledger and reconciliation to a measured total are materially shaped by accounting discipline.
- Operations Research — Bottleneck and process decomposition contribute channel definitions and recoverable-value analysis.
- Organizational & Management Science — Continuous-improvement ownership and loss accounting materially generalize the decomposition across operations.
Review resolution: Both independent reviews assign primary provenance to engineering_design. The queued secondary differences (alternate_origin_disagreement) are reconciled by retaining accounting_auditing, operations_research, organizational_management only as formative or independently established lineage(s), not merely as application domains. origin_mode=cross_disciplinary_synthesis records the provenance relationship, while domain_reach=multi_domain separately records applicability breadth. confidence=high preserves the more cautious assessment, and encyclopedia_synthesis=false records whether either reviewer identified a corpus-specific synthesis.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Overall Equipment Effectiveness is a manufacturing framework that decomposes the gap between theoretical and actual output into a fixed set of loss categories (the "six big losses" — breakdowns, setup, minor stops, speed loss, defects, startup). Its discipline of forcing losses to reconcile to a measured total is the model for any honest loss-channel account. ↩