← All conjectures · Texts, scribes & transmission
Benford audits Knossos
Status: Supported
The verdict’s fine print
— quoted from the resolution record:
“Pre-check passes everywhere (counted span 2.62-4.28 orders >= 2).”
Read the full caveats ↓
Status is derived only from the shepherd-authored triage/prediction data above -- community
submissions and claims are a separate overlay and can never change it (see the participation
panel below).
This is a conjecture imagined by a language model — drawn from its trained weights and held to falsifiability, novelty, and plausibility, not to any one method: it may join two or more fields, or none. It is
not an article and not evidence: it sits below the evidence/publication
boundary. A quantitative prediction and a named kill-dataset are attached (when registered)
so the claim stays falsifiable rather than merely evocative.
Claim (verbatim)
Benford audits Knossos. Benford's law — the striking regularity that genuinely counted quantities begin with the digit 1 about six times as often as with 9 — is here turned on the Linear B accounting tablets of Mycenaean Knossos. Numbers that arise from real enumeration of sheep, rations, and stores span orders of magnitude and emerge from multiplicative processes, so they should inherit the Benford first-digit distribution. Numbers set by administrative fiat — assessment quotas, targets, levies — are chosen for roundness and convenience, and should deviate from it. The conjecture is that this difference is detectable series by series: tablet series recording genuine tallies will be Benford-conformal, while series recording assessments will not. Digit statistics alone would thus separate what the palace actually counted from what it merely demanded.
Prediction clause (verbatim)
For each tablet series in the digit distributions by tablet series (the Knossos Linear B corpus grouped by series), extract all numerals and compute the first-digit distribution, testing Benford conformance with a mean-absolute-deviation (MAD) statistic; classify series independently as tally-type or assessment-type on epigraphic grounds. Primary clause: tally-type series conform to Benford (MAD below 0.015, Nigrini's marginal-conformance bound) while assessment-type series exceed it, in at least two-thirds of the series in each class. The verdict follows the primary clause.
Kill-dataset (verbatim)
digit distributions by tablet series.
In the kill-dataset registry:
On Inferpedia
This conjecture is linked to the following pages on Inferpedia, an encyclopedia of the missing — working atlas pages, some still early scaffolding.
Provenance
Run: Imported conversation (verbatim harvest)
· model: claude-fable-5
Origin: operator conversation with Claude Fable 5 at max effort, conducted 2026-07-03, relayed verbatim by the operator into the shepherd session on 2026-07-04. No ModelRun exists for the original generation (it happened outside the pipeline); this transcript file is the canonical capture. Transcript path: docs/generated/conjecture_harvest_fablemax_20260703.md. Model (operator-attested, not pipeline-recorded): claude-fable-5. Novelty disclaimer (verbatim, load-bearing -- rule 4): "Same caveat as before, doubled: at 100 items across all of archaeology and history, some of these will have cousins in the literature I can't check. What I can guarantee is the format — each links two things not normally linked, and each names the dataset or measurement that would kill it."
Novelty / leakage triage
no prior formulation located (search dated 2026-07-04)
Benford's law is standard in modern fraud/audit contexts, but no application to Linear B (or comparable Aegean/Near Eastern administrative tallies) distinguishing counted records from assessed quotas was located. No prior formulation located (search dated 2026-07-04). Methodological risk flagged for any future resolution: Linear B tallies are small counts spanning few orders of magnitude, so Benford conformance may be structurally unattainable regardless of honesty — a resolution must pre-check magnitude span or return inconclusive.
Sources cited by the triage
Predictions
Supported
registered 2026-07-04
Resolution: Supported
Caveats: Pre-check passes everywhere (counted span 2.62-4.28 orders >= 2). SUPPORTED on the pinned precedence: primary Pylos V1 significant (Delta=+0.0644, p=0.0314) with direction robust across ALL NINE cells (no variant flip); secondary Knossos decisive (V1 p<0.0001, V3 p=0.0006). Disclosed limits: Pylos V2 (p=0.172, assessment n=99 underpowered) and V3 (p=0.082) are directionally consistent but not individually significant; Knossos V2 sits under the pinned <40-numerals class guard (n=37); 84 compound-token digit runs (~1.5%) counted per the literal pinned regex (flagged, not repaired); 31 content-null docs all fall in excluded series; the task brief's '3,857' was a crawl-log tail misread -- the full 5,913-doc corpus was used. KILLED clause (assessment as-well-or-better) is false in every cell.
OPEN prediction (no in-house data): genuine counted Linear B tallies are closer to Benford-conformal than assessment/quota records, statistically separating counting from demanding within the same archive.
Resolution criteria — the registered fine print
Resolution criteria: Pre-check (mandated by triage): if the magnitude span of the tally corpus covers fewer than 2 orders of magnitude, the resolution is INCONCLUSIVE BY DESIGN (Benford not applicable to narrow-span counts). If applicable: SUPPORTED if first-digit distributions of counted series are significantly closer to Benford (chi-square distance, permutation p < 0.05) than quota/assessment series within the same tablet corpus. KILLED if quota series conform as well or better.
Known-priors disclosure — what the registrant already knew
Known priors disclosure: No in-house Linear B data. The registrant flagged at triage that small-count tallies may fail the span pre-check — this prediction may well resolve inconclusive-by-design, and that outcome must be reported as such, not as support.
Method and dataset — how it was measured
Register-before-compute: BLIND_SPEC.md + OPERATIONALIZATION.md staged before the crawl completed. Firewalled blind compute (claude-sonnet-5) under neutral class labels: per scope x variant, first-digit chi-square distance to Benford per class, Delta = D(assessment) - D(counted), one-sided permutation test (10,000 perms, seed 20260721, disclosed). Orchestrator spot-check reproduced the Pylos and Knossos V1 cells BYTE-IDENTICALLY via a fully independent implementation. Verdict claude-fable-5 per the pinned clause precedence (pre-check -> primary Pylos, V1 principal, direction-robustness across V1/V2/V3).
Dataset: Complete DAMOS corpus (Database of Mycenaean at Oslo), 5,913 published Linear B documents fetched 2026-07-20 via the site's own public JSON API (CC BY-NC-SA 4.0, robots 404, rate-limited, identifying UA -- lesson-4 permission verified FIRST), crawl ok=5,913 err=0. Classes pinned pre-compute from Bennett series letters (Docs^2/Chadwick): COUNTED={A,C,D,S,L} vs ASSESSMENT={M,N}; scopes Pylos (primary), Knossos (secondary), pooled (tertiary); three pinned tokenizer variants.
computed 2026-07-21
Weigh in
No community feedback yet.
New here? Create an account first
Create an account or sign in
and your feedback is tied to you — you can track it, get replies, and claim this conjecture
so others know you’re working on it. Prefer not to? Just leave your take below as a guest
— only the name you type is shown.
Add your take
Posted immediately (spam is removed). Community feedback is never an adjudicated verdict and
never changes this conjecture's triage label or status above.