Search Sensitivity Matrix¶
Template — instantiates Null Finding Warrant Calibration
Compares target forms against observation channels to show what each channel could and could not detect.
"We searched" hides a plural. A search is usually several channels — each looking for several possible forms of the target — and its blind spots live in the pairings, not in the whole. The Search Sensitivity Matrix is a two-dimensional template that makes those pairings explicit: target forms run down the side, observation channels run across the top, and each cell answers one question — would this channel catch this form, and at what level? Its defining idea is decomposition by pairing: it converts a single "did we find it?" into a grid of per-channel, per-form detection judgments, and the point of the exercise is to expose the rows no channel covers — the forms that would slip through every observer at once. Unlike a linear pass/fail review of the search as a whole, the matrix shows which form escapes which channel, and where those escapes overlap into a true blind spot.
Example¶
A contract manufacturer ships a precision casting and wants to state "no defects." The matrix lists the target forms it actually cares about as rows — surface scratch, subsurface void, dimensional drift, and inclusion of foreign material — and its inspection channels as columns: operator visual check, automated optical scan, coordinate-measuring machine, and ultrasonic testing on a one-in-fifty sample. Filling the cells is sobering. Surface scratches are caught well by two channels; dimensional drift only by the CMM; foreign inclusions only by ultrasonic. And subsurface voids fall through every channel except the ultrasonic sample — which sees only two percent of parts. The blank row for intermittent voids is the finding: a "no defects detected" claim covers scratches and drift confidently, but for voids it means only "none in the two percent we scanned." The matrix has turned a blanket reassurance into a map of exactly what the shipment's silence does and does not cover.
How it works¶
The template is built and read in a fixed rhythm:
- Enumerate the forms. List the distinct ways the target could actually appear — not "a defect" but the specific defect morphologies, each of which stresses detection differently.
- Enumerate the channels with their conditions. Each column is an observation channel annotated with its operating conditions: sampling fraction, threshold, timing, coverage.
- Fill each cell as a capability judgment. Mark detect / partial / miss for each form-channel pair, ideally with the level or rate at which detection holds — a cell is a small detection-power claim, not a checkmark.
- Read the empty cells and empty rows. A blank cell is one present-but-unseen scenario; a blank row is a form no channel catches — the search's genuine blind spot, which no amount of running the same channels harder will fix.
Tuning parameters¶
- Grid resolution — how finely forms and channels are split. Finer rows surface subtle blind spots but multiply cells and invite false precision.
- Cell content — binary detect/miss, a graded scale, or an explicit detection level. Richer cells are more honest and more work to justify.
- Condition annotation depth — how much of each channel's timing, threshold, and sampling is recorded in the column header versus assumed.
- Validation source — whether cells are filled from intuition, spec sheets, or measured performance against known positives; the last is far more trustworthy and far more costly.
When it helps, and when it misleads¶
Its strength is finding the blind row — the target form that no channel in the whole apparatus can see — which a channel-by-channel review taken one at a time will miss entirely. It also makes a "we searched thoroughly" claim inspectable cell by cell.
Its failure mode is false comprehensiveness: a fully populated grid looks exhaustive, and optimistic cells (marking "detect" where a channel only sometimes detects) quietly launder a weak search into a strong-looking one. Each cell is really a hit-rate under noise — the discipline of signal detection theory[n1] — so a cell filled from hope rather than from performance against known positives is the whole error in miniature. The guarding discipline is to calibrate cells against seeded or historical positives, not against what the channel is supposed to catch, and to treat any blank row as a warrant-limiting fact regardless of how full the rest of the grid looks.
How it implements the components¶
The matrix realizes the per-channel detection-mapping side of the archetype:
detection_power_model— each cell is a per-channel, per-form detection-capability judgment; the grid as a whole is the search's decomposed detection model.observation_condition_record— every column header records the channel and the conditions (sampling, threshold, timing) under which its cells hold.non_detection_alternative_set— the blank cells and blank rows enumerate exactly the present-but-unseen scenarios: which forms escape which channels.
It maps capability inside a search but does not set the search's outer extent or route to action: search_scope_boundary and follow_up_or_stopping_rule belong to its nearest twin, the Detection Power Checklist (which asks whether the whole search cleared each hurdle rather than decomposing channel by form) and to the Coverage Map and Blind-Spot Review.
Related¶
- Instantiates: Null Finding Warrant Calibration — the matrix supplies the per-channel detection picture the warrant depends on.
- Sibling mechanisms: Null Finding Warrant Memo · Detection Power Checklist · Negative Test Interpretation Protocol · Likelihood Ratio for Non-Detection · Minimum Detectable Presence Table · Coverage Map and Blind-Spot Review · Silent Monitor Assurance Review
Editorial Notes¶
Form Classification¶
Form family: Representation, Specification & Plan
Rationale: Search Sensitivity Matrix operates as a static representation, map, specification, schema, or prospective plan that externalizes information because it compares target forms against observation channels to show what each channel could and could not detect.
Independent corroboration: The frozen evidence defines Search Sensitivity Matrix as 'Compares target forms against observation channels to show what each channel could and could not detect', so its operative form is Representation, Specification & Plan.
Nearest alternative: Interface, Display & Cue — Search Sensitivity Matrix includes features of a user-facing prompt, display, template, or perceptual cue that shapes attention and action at the point of use, but its defining operation is a static representation, map, specification, schema, or prospective plan that externalizes information.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Convergent development
Present-day reach: Multi-domain
Rationale: Varying search inputs and tabulating output changes is sensitivity analysis expressed as a matrix. NIST and NASA define sensitivity analysis around quantifying how model outputs respond to input variation; engineering supplies the specific search design.
Related originating lineages:
- Data Science & Analytics — Data science, analytics, and operational monitoring supplies a parallel or contributing lineage for the mechanism's defining operation: compares target forms against observation channels to show what each channel could and could not detect.
- Engineering & Design — engineering_design contributes reliability, instrumentation, tolerances, verification, and systems integration to this mechanism's defining operation—Compares target forms against observation channels to show what each channel could and could not detect—without displacing the selected primary historical lineage.
- Information Theory — information_theory contributes signal, uncertainty, discrimination, and information loss to this mechanism's defining operation—Compares target forms against observation channels to show what each channel could and could not detect—without displacing the selected primary historical lineage.
- Mathematics — Mathematical modeling, proof, and abstract-structure practice supplies a parallel or contributing lineage for the mechanism's defining operation: compares target forms against observation channels to show what each channel could and could not detect.
Review resolution: The blind reviewers disagree on primary lineage (engineering_design versus statistics_experimental_design). Authoritative or primary research supports statistics_experimental_design as the best historical origin: Varying search inputs and tabulating output changes is sensitivity analysis expressed as a matrix. NIST and NASA define sensitivity analysis around quantifying how model outputs respond to input variation; engineering supplies the specific search design. The cited NIST, Guide for the Use of the International System of Units: Model Sensitivity; NASA, Sensitivity Analysis Overview directly supports the mechanism's defining operation. All independently supported contributing domains are retained without an arbitrary cap. origin_mode=convergent records the lineage relationship, while domain_reach=multi_domain records later applicability separately from provenance.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
- NIST, Guide for the Use of the International System of Units: Model Sensitivity
- NASA, Sensitivity Analysis Overview
Notes¶
[n1] Signal detection theory treats every detection as a decision under noise, with a hit rate and a false-alarm rate that trade off along an operating characteristic. It is why a matrix cell should carry a rate, not a yes/no: a channel that "can" see a form still misses it some fraction of the time. ↩