Exact-N Count Audit¶
Counting protocol — instantiates Claim Quantifier Scope Calibration
Verifies a cardinality claim by fixing what counts as one unit and running an exhaustive census against that basis.
Some quantifiers are not vague at all — they name a number. "There are exactly eight backup copies." "We support at least three regions." "Only two people hold this key." Exact-N Count Audit is the protocol for claims of cardinality, and its defining move is that it settles the unit of count before it counts. Most cardinality errors are not arithmetic mistakes; they are disagreements about what "one" means — one copy or one copy-set, one region or one availability zone, one person or one role. The audit fixes the counting basis, states the exact predicate an item must satisfy to be tallied, and then performs a census rather than a sample, because "exactly N" is a claim that admits no rounding.
Example¶
A compliance reviewer is handed the assertion: "We keep exactly three offsite backups of the customer database." Before counting anything, the audit fixes the basis. What is one backup — a full snapshot, or does an incremental chain count as one? Is a copy that failed its last integrity check still a backup? The predicate is stated: a backup is a full, restore-verified snapshot stored outside the primary region within the last 24 hours. Now the census runs against that predicate: the reviewer enumerates every candidate copy and checks each. Two full snapshots pass. A third exists but failed verification — it does not satisfy the predicate. A fourth turns out to be in the same region and fails the "offsite" clause. The count is not three; it is two. The claim was false not because someone miscounted but because "backup" had silently included copies the predicate excludes. The audit's output is the corrected token — exactly two — plus the basis that makes it checkable.
How it works¶
- Define the unit. State what constitutes a single countable item, resolving the "one copy vs. one copy-set" ambiguity before any tally begins.
- State the counting predicate. Fix the exact property an item must satisfy to be included, so borderline items are decided by rule, not by mood.
- Census, don't sample. Enumerate the whole domain and test each candidate against the predicate; an exact-N claim cannot be established by a sample the way a "most" claim can.
- Watch the fenceposts. Reconcile the tally against off-by-one traps — inclusive vs. exclusive bounds, duplicates counted twice, items straddling two categories.
Tuning parameters¶
- Unit granularity — how finely a "one" is defined. Coarse units are quick to count but hide internal multiplicity; fine units are precise but multiply the census work.
- Predicate strictness — how demanding the inclusion test is. A strict predicate (restore-verified, in-window) yields a defensible count but rejects items a lax reader would have tallied.
- Census completeness — full enumeration versus a bounded sweep with a documented cutoff. Anything short of full enumeration converts an exact claim into an estimate and should be labelled as one.
- Reconciliation rigor — how hard you hunt for double-counts and fencepost errors before certifying the number.
When it helps, and when it misleads¶
Its strength is that it makes a number auditable: anyone can rerun the census against the stated basis and get the same N, and disputes move from "is it three?" to the sharper "does this copy satisfy the predicate?" It is the natural home for fencepost discipline[n1], where most exact-count errors actually live.
Its failure mode is that an exact count is only as stable as its basis, and a basis can be gamed. Loosen the predicate and the number climbs; tighten it and the number falls — so a motivated auditor can hit almost any target by tuning the unit definition rather than miscounting. The classic misuse is quoting the headline N while quietly swapping the basis that produced it ("three backups" under a definition that counts unverified copies). The guarding discipline is to publish the counting basis alongside the number and to freeze it before the count runs, so the predicate cannot be reverse-engineered from the answer you wanted.
How it implements the components¶
measure_or_count_basis— its foundational product: the explicit unit-of-count and inclusion rule the number is measured against.quantifier_token_record— it records the exact cardinality token ("exactly two," "at least three") that the census establishes.predicate_under_quantification— it pins down the precise property each counted item must satisfy.
It does not test whether the count crosses a majority threshold_statement, nor does it document a scope_shift_log — those belong to the most-threshold statement and domain checklist. Its nearest twin is the negative-claim exhaustion check, which also enumerates a domain; but that one certifies an exhaustion_record proving a count of zero, whereas this audit certifies an exact positive N.
Related¶
- Instantiates: Claim Quantifier Scope Calibration — supplies the cardinality-verification step of the appraisal.
- Consumes: Quantified Claim Template supplies the quantifier token and predicate the census then verifies.
- Sibling mechanisms: Negative-Claim Exhaustion Check · Existential Witness Card · Most-Threshold Statement
Editorial Notes¶
Form Classification¶
Form family: Assessment, Review & Assurance
Rationale: Exact-N Count Audit operates as a bounded evaluation of existing evidence or work that produces a finding or disposition because it verifies a cardinality claim by fixing what counts as one unit and running an exhaustive census against that basis.
Independent corroboration: The frozen evidence defines Exact-N Count Audit as 'Verifies a cardinality claim by fixing what counts as one unit and running an exhaustive census against that basis', so its operative form is Assessment, Review & Assurance.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Universal
Rationale: Defining units and performing an exhaustive census are core statistical enumeration and measurement practices.
Related originating lineages:
- Accounting & Auditing — Audit traditions materially contribute independent verification of asserted counts against a fixed counting basis. Inventory and completeness audits independently developed controlled count procedures and reconciliation to a defined population.
Review resolution: Both reviewers agree that statistics_experimental_design is primary. I retain accounting_auditing only as formative origin lineages; cross_disciplinary_synthesis is appropriate because the final form materially combines the agreed primary with the retained formative lineages. Reach is universal because the structure is portable across essentially any domain with the stated problem, an applicability judgment kept separate from provenance. Encyclopedia synthesis is true because the exact generalized artifact is an encyclopedia-authored combination or refinement. No unresolved historical ambiguity remains after reconciling the secondary fields.
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] A fencepost (off-by-one) error miscounts by confusing the number of segments with the number of endpoints — the reason "how many posts in a 10-metre fence with posts every metre" trips people up. Exact-N audits fail here far more often than in the arithmetic itself. ↩