Pairwise Dominance Audit¶
Test or assessment — instantiates Cyclic Dominance Counterbalancing
A review procedure that tests whether pairwise comparisons form a genuine cycle, a simple ranking, noise, or context-specific dominance.
A Pairwise Dominance Audit is the evidence check that asks whether a claimed cycle deserves to be governed as one. It re-examines the repertoire, holds each pairwise edge up to a confidence bar, and returns one of four verdicts — a genuine intransitive cycle, a plain transitive ranking wearing a costume, measurement noise, or context-specific dominance that only looks cyclic when contexts are mixed. Its defining move is that it is skeptical by construction: it assumes the cycle might be an artifact until the evidence clears the bar, records the verdict as a durable legitimacy record, and — where the answer is "no real cycle" — fires the exit condition so the organization stops paying to counterbalance a phantom.
Example¶
A hospital's stewardship committee is asked to adopt an antibiotic cycling protocol on the theory that its three drug classes form a collateral-sensitivity loop: bacteria that evolve resistance to class A become newly vulnerable to B, resistance to B opens a weakness to C, and resistance to C loops back to A. Before committing wards to a rotation, the audit tests the loop. Are the reversals reproducible across many bacterial isolates, or driven by one unusual strain? Are the pairwise effects large enough to distinguish from chance once you account for the fact that many drug pairs were compared? The audit finds two edges hold up but the third is indistinguishable from noise — the loop does not close. Verdict: not a governable cycle. It records exactly what was tested and why, and fires the break condition — do not run this as a rock-paper-scissors rotation on the strength of two-and-a-half edges. The ward avoids building a protocol on a pattern that would have dissolved under a wider sample.
How it works¶
- Re-audit the repertoire. Confirm the option set is clean — no decoys, no duplicated options, no permanently dominated ones — because a phantom option can manufacture a phantom edge.
- Set an evidence bar per edge. Each pairwise claim must clear a confidence and effect-size threshold before it counts as a real edge rather than a fluke.
- Correct for multiplicity. When many pairs were compared, some will look cyclic by chance; the audit adjusts the bar for how many comparisons were run.
- Classify the pattern. Sort the result into cycle / ranking / noise / context-specific, and — when it is not a stable cycle — fire the exit condition rather than defaulting to counterbalancing.
Tuning parameters¶
- Confidence bar per edge — how much evidence an edge needs to be drawn. A high bar suppresses phantom loops but can erase a real, faint cycle.
- Multiplicity correction — how aggressively the bar is raised for the number of comparisons. Strong correction guards against spurious loops but risks missing weak true ones.
- Replication requirement — how many independent samples or episodes an edge must survive. More replication is more trustworthy but slower and costlier.
- Classification thresholds — how strong the pattern must be before it is called a "cycle" versus "context-specific." Loose thresholds over-diagnose cycles.
- Audit cadence — one-time certification versus periodic re-audit as the field adapts and edges shift.
When it helps, and when it misleads¶
Its strength is that it kills false cycles before they cost anything — it is the gate that stops a team from imposing a rock-paper-scissors story on what is really a ranking with noise, and the durable record it leaves makes the whole governance decision reviewable instead of a matter of taste.
Its failure mode cuts both ways. An underpowered audit can miss a real but faint cycle and wave it off as noise; and an audit run too eagerly can manufacture a cycle by slicing the data until three favorable edges appear — the look-elsewhere effect, where enough comparisons guarantee some cyclic-looking pattern by chance.[n1] The classic misuse is exactly that p-hacking: testing pairs until a satisfying loop emerges, then declaring the cycle "audited." The guarding discipline is to pre-commit to the edges and thresholds before looking, correct honestly for how many comparisons were run, and require replication before a cycle is certified as governable.
How it implements the components¶
option_repertoire— the audit re-validates which options genuinely belong in the set, since a mis-specified repertoire is the commonest source of a fake edge.cycle_legitimacy_record— its output is the durable, reviewable record of what was tested, what cleared the bar, and the verdict that certifies (or denies) the cycle.cycle_break_condition— a "not a real cycle" or "one option now dominates" verdict fires the condition to exit the cyclic frame rather than keep counterbalancing.
It does not produce the relation data it scrutinizes — those magnitudes and directions come from Cyclic Payoff Table and Beats-Relation Matrix — and it neither maintains options nor rotates them; the floors belong to Portfolio Minimum-Viability Rule and the activation logic to Countermove Rotation Playbook.
Related¶
- Instantiates: Cyclic Dominance Counterbalancing — it is the evidence gate that keeps the whole apparatus from governing a cycle that isn't there.
- Consumes: Cyclic Payoff Table and Beats-Relation Matrix — it audits the claims those artifacts record.
- Sibling mechanisms: Beats-Relation Matrix · Cyclic Payoff Table · Countermove Rotation Playbook · Portfolio Minimum-Viability Rule · Adaptive Mix Review · Sentinel Option Trial · Nontransitive Scenario Simulation
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Pairwise Dominance Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it a review procedure that tests whether pairwise comparisons form a genuine cycle, a simple ranking, noise, or context-specific dominance.
Independent corroboration: The frozen evidence defines Pairwise Dominance Audit as 'A review procedure that tests whether pairwise comparisons form a genuine cycle, a simple ranking, noise, or context-specific dominance', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Economics & Finance
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Pairwise Dominance Audit is most directly rooted in economics and finance's analysis of scarcity, incentives, tradeoffs, contracts, and option value. The lineage fits its defining practice: A review procedure that tests whether pairwise comparisons form a genuine cycle, a simple ranking, noise, or context-specific dominance.
Related originating lineages:
- Mathematics — Pairwise Dominance Audit also draws materially on mathematics' axiomatic study of abstract structure, relations, and formal operations, which shaped this mechanism rather than merely adopting it as an application.
- Operations Research — Pairwise Dominance Audit also draws materially on operations research's mathematical optimization, simulation, queues, decision analysis, and resource allocation, which shaped this mechanism rather than merely adopting it as an application.
- Statistics & Experimental Design — Multiple-comparison correction and replication materially shape whether an apparent cycle exceeds noise.
Review resolution: Both independent reviews agree on primary origin economics_finance; reconciliation resolves reported_ambiguity, alternate_origin_disagreement. Formative alternate lineages retained: mathematics, operations_research, statistics_experimental_design. The broader reach of later applications is kept separate as domain_reach=multi_domain; origin_mode=cross_disciplinary_synthesis records how the formative lineages relate. Confidence is conservatively reconciled to medium, and encyclopedia_synthesis=true preserves the reviewers' boundary judgment.
Attribution caveat: The audit combines preference theory with statistical testing and does not belong wholly to one lineage.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] The look-elsewhere effect (a face of the multiple-comparisons problem) — when many hypotheses are tested at once, some will appear "significant" by chance alone, so an uncorrected search almost always turns up a striking-looking pattern. Applied here, testing enough option pairs will surface a cyclic-looking loop even in noise, which is exactly why the audit must correct for the number of comparisons before certifying a cycle. ↩