Linear A, the undeciphered writing system of Minoan Crete (c. 1800–1450 BCE), has resisted a century of attempts to recover the language behind it. This audit does not attempt decipherment. It treats the surviving clay tablet corpus as a mute, non-linguistic inventory database and asks a question that does not require knowing what the signs sound like: does the arithmetic close?
The paper names the failure mode it is avoiding. Mechanically projecting phonetic values from the deciphered Linear B script onto homomorphic Linear A signs is, in its terms, a sycophancy failure — an open loop in which the output is never tested against anything that could contradict it. The same structural fault this author audits in language models, performed on clay.
| Named protocol | Projection-before-meaning — mathematical invariants constrain all semantic interpretation |
|---|---|
| Rule 1 | The reward function is continuously verified against the mathematical identities and balance equations of the tablets |
| Rule 2 | Any proposed sign reading or phonetic mapping that violates a verified sum, fractional closure or deficit equation is automatically rejected |
| Rule 3 | The quantitative layer is processed first; the qualitative layer is constrained to solutions consistent with verified arithmetic |
| Corpus | GORILA (Godart & Olivier, 1976–1985), with the "Vase" volumetric simulation (Younger, 2003) and ceramic analysis (Best, 1982) |
| Bibliography | 135 cited sources |
| Does not attempt | Phonetic values, language family identification, or readable translation |
The point of the constraint is that it can fail. A reading that violates a closed sum is discarded regardless of how attractive it is linguistically — which means the arithmetic, not the interpreter, decides what survives.
Deficits behave consistently across the corpus: 5 units in the fluid columns of HT 94, 197 on the personnel register HT 1. The same operator, two orders of magnitude apart, in two different commodity classes.
| Vessel | Declared | Physical capacity | Result |
|---|---|---|---|
| ZA Zb 3 | 32 units = 921.6 L | ~556 L usable (996 L external) | +365.6 L · 65.7% |
| KN Zb 35 | 100 units oil = 2,880 L | 72–292 L per pithos | requires 10–15 vessels |
| KN Zb 27 | 117 units = 3,369.6 L | ~150–250 L | order of magnitude over |
ZA Zb 3 stands 1.7 m with walls of 1.7–4.5 cm; the discrepancy is not a measurement artefact. KN Zb 27 was read as 17 units for decades because the pithos itself had been mislaid in museum storage and the transcription rested on archival drawings; the corrected reading is 117.
Across the three vessels, declared volumes exceed physical capacity by 66% to over 1,000%. The audit's reading follows from that: these are warehouse manifests, not container labels — the jar carries the record for a consignment distributed across many vessels, exactly as a modern pallet label does.
| ku-ro | Vertical aggregation operator — sums partial entries within a single ledger column or tablet section |
|---|---|
| ki-ro | Negative-balance operator — a missing delivery, transport shrinkage, tax debt or receivable carried to the next period |
| sa-ra2 | Header operator — classifies resources as palatial reserve or central warehouse stock |
| po-to-ku-ro | Higher-order summation — consolidates partial totals across sections, tablet faces or territorial units |
The audit also reclassifies the so-called libation formula — long read as ritual text — as a rigid six-field stock-keeping record. Whether the underlying words were Semitic, Hurrian, Kartvelian or Indo-European is treated as secondary to their verified computational behaviour, which is the same position the protocol takes throughout.
Each prediction names the evidence that would destroy it and the count at which it is destroyed. That is the difference the paper claims between a testable structural claim and an interpretive proposal.
Nothing here recovers Minoan. The etymological affiliation of the operators — Semitic, Hurrian, Kartvelian or Indo-European — is explicitly treated as secondary to their verified computational behaviour, not as solved.
The paper describes itself as a testable structural claim rather than an interpretive proposal, and makes no claim about which position a reader should adopt regarding how much evidential weight tablet arithmetic carries.
The functional definitions of the operators, the reclassification of the libation formula as a six-field record, and the reading of the inscribed vessels as regional manifests are analytical inferences drawn from ledger layout and balance equations — not counted quantities.
Diener, Ł. (2026). Excel in Clay I: The Minoan Pipeline — A Non-Linguistic Structural Audit of Linear A as an Integrated Supply Chain Database (version 1.0). Zenodo. https://doi.org/10.5281/zenodo.20442145
Preprint, not peer reviewed by a journal. Licensed CC BY 4.0. First of three audits applying the same protocol to undeciphered Bronze Age administrative corpora; the numbering is thematic, and publication order was III, I, II.
Read on Zenodo Excel in Clay II Excel in Clay III All six audits