CLM-0014 — Exact word-count equalities are evidence of design
REFUTED_AS_STATED Audit report · deterministic reproduction
1 · Claim as commonly stated
When two meaningfully paired words occur exactly the same number of times, that equality is evidence of deliberate design rather than coincidence.
Source note: The implicit premise beneath the numerical-miracle genre as a whole; not attributable to one author.
2 · Reproduction method
Declared counting-rule matrix, executed against corpus quran:hafs-kufan:v1. Full rules, counts, and sample locations: EXP-0012.
3 · Result
Tested directly in EXP-0012 by measuring how much raw material an equality claim has to draw on. Among Qur'anic lemmas occurring 20+ times, 1,640 pairs share an exact count (1.3% of all pairs). The same measurement on Sahih al-Bukhari gives 40,912 pairs (1.5%) and on al-Mutanabbi's Diwan 14 pairs (2.2%) — the same order of magnitude in every text.
4 · Sensitivity to counting rules
REFUTED_AS_STATED — exact equality between two word counts is an ordinary feature of any Zipf-distributed vocabulary, not a distinguishing property of any one text.
5 · Statistical context
The premise fails for post-hoc pairs and survives only for pairs named BEFORE counting, where the base rates measured here (0.7%–4.5% depending on frequency band) would apply as a genuine prior.
6 · Best argument in favor
For a pair specified in advance, an exact tie really is unlikely — the base rates here quantify how unlikely, and they are low.
7 · Strongest objection
No circulating claim was specified in advance; each was found by searching, and the searchable stock of ties is in the thousands.
8 · All recorded objections
- No circulating claim was specified in advance; each was found by searching, and the searchable stock of ties is in the thousands.
- Comparison texts show the same rate, so the phenomenon is not a property of the Qur'an.
9 · Final status
REFUTED_AS_STATED
10 · Public summary
Exact ties between word counts are common in every book. They only carry weight when the pair is named before anyone counts.
11 · Reproduce it yourself
python3 services/research-worker/scripts/run_experiment.py data/research/specs/EXP-0012.json — the registry entry lives in data/research/claims.json; the result files record input checksums and environment.