Counterexample Register¶
Falsification register — instantiates Predicate Criterion Formalization
Keeps a running log of the cases that falsify or strain a criterion, turning refutations into the trigger for revising it.
A Counterexample Register is a running log of the cases a criterion gets wrong or can barely handle — the false positives, false negatives, and borderline specimens that strain or falsify it — kept specifically to drive the criterion's revision. Its defining orientation is refutation, not confirmation: where a validation suite asks whether the predicate still passes its blessed cases, the register hunts for and preserves the cases that break it, treating each as evidence that the definition may need to change. Its two jobs are to read the accumulation as a signal that the criterion is drifting out of fit, and to give a refuting case a path to actually override or revise the rule rather than being quietly ignored.
Example¶
An epidemiology team maintains a case definition for a newly emerging illness — the predicate that decides whether a patient counts as a "case" for surveillance. As the outbreak evolves, clinicians submit patients who strain the definition: a lab-confirmed patient who lacks the required fever (a false negative under the current rule), and a symptom-matching patient who later tests negative (a false positive). Each goes into the counterexample register with the detail of how it broke the definition.
Individually these are anomalies; in aggregate the register becomes a drift signal — when confirmed-but-excluded cases pile up, that is evidence the fever requirement is now too strict for a mutated strain. Crucially, the register is not a graveyard: a threshold of accumulated counterexamples triggers a formal review, the override path by which the definition committee revisits and, here, relaxes the fever criterion. Setup to outcome: the messy stream of "the rule got this one wrong" stops being lost in individual judgment calls and becomes the organized pressure that keeps the criterion honest to reality.
How it works¶
- Capture refutations, not confirmations. Log the cases the criterion mishandles — false positives, false negatives, borderline and adversarial specimens — with how each strains the rule.
- Read the aggregate as drift. Track the accumulation and its pattern; a rising or clustering count is the signal that the criterion is losing fit with its domain.
- Route to override or revision. A refuting case, or a threshold of them, triggers a defined path to challenge and change the predicate — never a silent discard.
- Preserve the challenge to universals. A single counterexample suffices to refute a categorical claim; the register keeps it available to do so.
Tuning parameters¶
- Capture threshold — how strained a case must be to log. A low threshold catches early drift but adds noise; a high threshold logs only clear breaks.
- Trigger rule — how many, or what pattern of, counterexamples forces a review. Sensitive triggers revise often (churn); insensitive ones let staleness accumulate.
- Weighting — treating a single decisive counterexample (a true falsification) differently from many marginal strains.
- Provenance depth — how much context each case carries, trading capture cost against usefulness at revision time.
When it helps, and when it misleads¶
Its strength is that it keeps a criterion answerable to reality: it operationalizes falsifiability[n1] by making refutations first-class, and converts scattered "the rule blew this one" complaints into organized pressure for change with an actual path to change.
Its failure mode is register capture — the log fills with cases but no trigger ever fires, so counterexamples accumulate as folklore while the stale definition stands. The opposite misuse is over-reaction: revising the criterion on a single unrepresentative case and destabilizing it. The discipline that guards against both is to define in advance what accumulation triggers a review and to weight decisive falsifications above marginal strains, so the register drives revision without thrashing.
How it implements the components¶
drift_monitor— the accumulating counterexamples ARE the drift instrument: their count and pattern signal when the criterion has lost fit with its domain.appeal_or_override_path— a logged refutation, or a threshold of them, is granted a defined route to challenge, override, and revise the predicate rather than being discarded.
It does NOT confirm the criterion against a fixed grid of expected verdicts — the boundary_case_library and implementation_trace of that regression role are Test Case Matrix; and it does not store the successive rule versions a refutation eventually produces — the predicate_owner_and_change_log ledger is Predicate Version Registry.
Related¶
- Instantiates: Predicate Criterion Formalization — the register is the refutation-and-revision instance, keeping the criterion under pressure from the cases it gets wrong.
- Sibling mechanisms: Boolean Guard Clause · SQL WHERE Clause or Query Filter · Truth Table · Decision Table · Eligibility Criteria Checklist · Policy Definition of Terms · Predicate Version Registry · Test Case Matrix · Unknown-State Routing Rule
Editorial Notes¶
Form Classification¶
Form family: Record, Log & Register
Rationale: Counterexample Register operates as a durable record, ledger, register, or trace whose value depends on preserving actual state or history because it keeps a running log of the cases that falsify or strain a criterion, turning refutations into the trigger for revising it.
Independent corroboration: The frozen evidence defines Counterexample Register as 'Keeps a running log of the cases that falsify or strain a criterion, turning refutations into the trigger for revising it', so its operative form is Record, Log & Register.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Philosophy
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Logic and philosophy of science cohered preserving refuting cases as privileged evidence against a predicate or general rule.
Related originating lineages:
- Data Science & Analytics — Error logging supplies structured false positives, false negatives, and borderline cases for model improvement.
- Organizational & Management Science — Issue-register practice supplies a persistent owner-reviewed ledger and an escalation path from accumulated exceptions to rule revision.
Review resolution: The running artifact synthesizes philosophical falsification with data exception logging and organizational follow-through; that operational packaging warrants the synthesis flag.
Attribution caveat: The register form operationalizes counterexample reasoning through quality and model-management practice.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Karl Popper's criterion of falsifiability: a claim earns scientific standing not by the confirmations it collects but by naming what would refute it. A counterexample register is that criterion made operational for a predicate — it treats a single decisive counterexample as more informative than a thousand passing cases, because the counterexample is what forces the definition to change. ↩