Resolution: Killed
Caveats: This conjecture, submitted by RMD, wagered that once a pre-modern polity passes 100,000 people, hereditary centralized rule becomes the overwhelming default — while most small-scale societies never build any hierarchy beyond the local community at all. It died on both blades, and the registration I wrote before seeing a single number leaves no room to argue: the Seshat share came in at 61/103 = 0.592, below the 0.60 kill line, and the ethnographic premise failed too, with 630 of 1,155 EA033-coded societies (54.5%) showing supra-local hierarchy under my blind binding that petty chiefdoms count. The genuinely interesting texture is in how it died. Half of the intuition is overwhelmingly right: hereditary elite status is near-universal among big polities — 102 of 103, 99% — and every polity that coded centralized also coded hereditary. What fails is unitary centralization: the large-polity record is full of empires that Seshat codes confederated, nominal, or loose — Carolingian, Seljuk, Parthian, Vijayanagara, Majapahit, the Buyids — scaled polities held together by something other than a unitary state. A defender could contest two of my own bindings: that only 'unitary'/'unitary state' counts as centralized (imperial confederations arguably still have a king at scale), and that EA033 code 2 — one level beyond local, petty chiefdoms — counts as kingship-relevant hierarchy. Both are fair fights, but they would require a new registration adjudicated blind, not a reinterpretation of this one; under the bindings as written, the conjecture is killed, unsoftened. GUARD WALK-THROUGH (adjudicator): SESHAT CLAUSE: denominator 103 unique population-qualified PolIDs coded on both variables; numerator 61; ratio 61/103 = 0.59223 exact. MINIMUM-N GUARD: 103 >= 30, does not fire. COVERAGE GUARD: 32/135 = 0.2370 excluded for UNKNOWN/missing, below the 0.50 trigger, does not fire. BOUNDARY GUARD: compared as exact fractions unrounded, 61/103 < 3/5 strictly (305 < 309), so the ratio falls in the kill band (<0.60), not the unresolved band [0.60,0.80). HEREDITARY-PROXY GUARD: the joint ratio is in the kill band, so the guard is engaged; it downgrades the kill to UNRESOLVED only if the centralized-only share on the same denominator is >= 0.80, attributing the shortfall to the proxy. Centralized-only = 61/103 = 0.59223, itself < 0.60, and the registration binds this case explicitly: the kill STANDS because the centralization instrument is direct. The proxy is not the failure point at all — hereditary-only is 102/103 = 0.99029, and every centralized polity also codes hereditary, so the entire shortfall sits on 'Degree of centralization' under the bound unitary/unitary-state positive set. Seshat kill fires and stands. D-PLACE CLAUSE: 1,155 EA033-coded societies; codes 2-5 = 630; supra_local_share = 630/1155 = 6/11 = 0.54545, strictly > 0.50 (1260 > 1155), so the independent D-PLACE kill fires under the blind code-2-counts-as-supra-local binding. D-PLACE SEQUENCING GUARD: engaged as disclosed; thresholds and the code-2 binding were authored blind to the marginals, so the clause verdict STANDS, with the sequencing disclosure reproduced verbatim in this ledger. DATA-VERSION GUARD: only the bound Equinox June 9 2022 xlsx sheets and the fixed dplace_EA_data.csv were used per the computation record; schemas matched; does not fire. NO-POST-HOC GUARD: no measurement ambiguity outside the registration arose during adjudication; nothing to resolve toward UNRESOLVED and no registration fault to log. COMPOSITE VERDICT: KILL = Seshat ratio < 0.60 OR supra_local_share > 0.50; both disjuncts fire independently and each survives its guard. Verdict: killed. SEQUENCING DISCLOSURE (verbatim, per the registration's own guard): the operator computed the EA033 marginal distribution before registration while checking data availability; the numbers were withheld from the registrar; thresholds were authored independently beforehand; the code-2 decision was bound blind by the registrar.
PRIMARY (binding): in the Seshat Global History Databank (Equinox release), at least 80% of pre-1500 CE polities whose peak polity population exceeds 100,000 code hereditary succession combined with centralized government, while fewer than 50% of societies coded on D-PLACE Ethnographic Atlas EA033 show any jurisdictional hierarchy beyond the local community. Confirm: Seshat >=0.80 AND D-PLACE clause holds. Kill: Seshat <0.60, or EA033 supra-local share >0.50. Unresolved band: Seshat [0.60,0.80).
Resolution criteria: SESHAT CLAUSE: SOURCE: Seshat Equinox release (June 9 2022 xlsx), sheets Equinox2020_CanonDat and Polities only. UNIT OF ANALYSIS: unique PolID from the Polities sheet; rows sharing a PolID across NGAs are deduplicated and the polity is counted exactly once. PRE-1500 RULE: a polity qualifies iff its Polities-sheet Start year is < 1500 (years are astronomical integers; negative = BCE, so all BCE polities qualify). FACT-WINDOW RULE (applied uniformly to all three variables below): a Equinox2020_CanonDat row for a qualifying polity is included iff it is undated, or Date.From < 1500, or (Date.From missing and Date.To < 1500). FACT.TYPE RULE: for each polity-variable pair, use rows with Fact.Type='simple'; if no simple rows exist, fall back to Fact.Type='complex' rows, and if the complex rows disagree on the binary classification defined below, the polity is UNKNOWN on that variable. PEAK POPULATION: variable 'Polity Population'; peak = the maximum over all numeric values parsed from both Value.From and Value.To across all included rows (values trimmed, commas and whitespace stripped before parsing; unparseable or non-numeric values ignored; if no numeric value survives, the polity is not population-qualified). A polity enters the population-qualified set iff peak > 100000 STRICTLY; a peak of exactly 100000 does not qualify. CENTRALIZED CLASSIFICATION: variable 'Degree of centralization', values matched case-insensitively on trimmed strings. Centralized set: 'unitary', 'unitary state'. Not-centralized set: 'confederate state', 'confederated state', 'confederated state (1792 BCE-)', 'confederation', 'loose', 'nominal', 'nominal allegiance', 'nominal(-1792 BCE)', 'none', 'quasi', 'quasi-polity'. Treated as UNKNOWN (non-classification-bearing): 'polity', 'unknown', 'suspected unknown', and any value string not listed here. HEREDITARY CLASSIFICATION (proxy variable, see guards): variable 'elite status is hereditary'. Hereditary set: 'present', 'inferred present'. Not-hereditary set: 'absent', 'inferred absent'. UNKNOWN: 'unknown', 'suspected unknown', and any unlisted value. ANY-FACT RULE (bound now, disclosed as the permissive direction): a polity CODES centralized (respectively hereditary) iff at least one included classification-bearing fact falls in the positive set; it codes not-centralized (respectively not-hereditary) iff it has at least one included classification-bearing fact and none in the positive set; it is UNKNOWN on the variable iff it has no included classification-bearing fact. DENOMINATOR: population-qualified polities that are non-UNKNOWN on BOTH 'Degree of centralization' AND 'elite status is hereditary'. Polities UNKNOWN or missing on either variable are excluded from the denominator, and their count MUST be reported alongside the result (coverage guard applies). NUMERATOR: denominator polities coding BOTH centralized AND hereditary. RATIO: numerator divided by denominator, kept as an exact fraction; no rounding before threshold comparison. D-PLACE CLAUSE: SOURCE: docs/generated/_community1_data/dplace_EA_data.csv (already fetched, fixed; no re-fetch), Ethnographic Atlas variable EA033 'Jurisdictional hierarchy beyond local community'. DENOMINATOR: all societies carrying a non-missing EA033 code in {1,2,3,4,5}, each society counted once by its society identifier; expected 1155, but the ACTUAL non-missing count in the fixed file governs. SUPRA-LOCAL SET (bound now, blind to the marginals): codes 2, 3, 4 and 5 ALL count as 'any jurisdictional hierarchy beyond the local community' — code 2 (one level beyond local, e.g. petty chiefdoms) EXPLICITLY COUNTS as supra-local. Define supra_local_share = count(codes 2 through 5) / denominator, as an exact fraction. CLAUSE HOLDS iff supra_local_share < 0.50 strictly (equivalently, share of code 1 > 0.50). KILL fires iff supra_local_share > 0.50 strictly (a majority of coded societies show supra-local hierarchy). supra_local_share exactly equal to 0.50: the clause neither holds nor kills; unresolved on this clause. ROLE IN COMPOSITE VERDICT: confirmation of the conjecture requires the Seshat ratio >= 0.80 AND this clause holding; this clause's kill fires independently of the Seshat result. BINDING GUARDS: - MINIMUM-N GUARD (Seshat): if the final denominator (population-qualified polities coded on both variables) contains fewer than 30 unique polities, the Seshat clause verdict is INCONCLUSIVE — neither confirm nor kill may fire from it — and the composite verdict is unresolved. - COVERAGE GUARD (Seshat): if more than 50% of the population-qualified polities are excluded from the denominator for UNKNOWN/missing coding on either variable, the Seshat clause verdict is INCONCLUSIVE regardless of the ratio; missingness that large can select the sample in either direction and supports neither confirm nor kill. - HEREDITARY-PROXY GUARD: 'elite status is hereditary' is bound as a PROXY for hereditary succession because the Equinox release carries no direct ruler-succession variable. If the joint ratio falls in the kill band (< 0.60) while the centralized-only share on the SAME denominator is >= 0.80 — i.e., the shortfall is attributable to the hereditary proxy — the kill DOWNGRADES to UNRESOLVED with an instrument-validity note; it does not stand. If the centralized-only share is itself < 0.60, the kill STANDS (the centralization instrument is direct). A confirmation reached via the proxy stands as CONFIRMED but must carry the proxy disclosure and the bias-direction note (instrument_notes) verbatim in the ledger. - D-PLACE SEQUENCING GUARD: the operator computed the EA033 marginal distribution before this registration (disclosed; numbers withheld from the registrar). The thresholds were authored before that computation, and the one remaining degree of freedom — whether code 2 counts as supra-local — is bound here by the registrar without sight of the marginals, in the only defensible direction (code 2 is by definition a level beyond the local community). The D-PLACE clause verdict therefore STANDS, but the resolution ledger must reproduce the sequencing disclosure verbatim next to the clause result. - BOUNDARY GUARD: all shares are exact fractions compared unrounded. Seshat: exactly 0.80 CONFIRMS the clause (>= is inclusive); exactly 0.60 is UNRESOLVED (kill is strict <0.60); the unresolved band is [0.60, 0.80). D-PLACE: both inequalities strict as bound; exactly 0.50 is unresolved. Rounding is never applied before a threshold comparison. - DATA-VERSION GUARD: Seshat measurement uses ONLY the Equinox June 9 2022 xlsx, sheets Equinox2020_CanonDat and Polities; D-PLACE uses ONLY the already-fetched docs/generated/_community1_data/dplace_EA_data.csv. No substitution of later releases, re-fetches, or hand-patched rows. If either file is unreadable or its schema does not match this registration, the affected clause is INCONCLUSIVE; the adjudicator may not improvise an alternative instrument. - NO-POST-HOC GUARD: any measurement ambiguity not bound in this registration must be resolved AGAINST whichever reading would flip the verdict band — interpretive ties go to UNRESOLVED — and every such gap must be logged in the ledger as a registration fault of the registrar, not adjudicated silently. INSTRUMENT NOTES: COMPOSITE VERDICT LOGIC (subject to guards): CONFIRM = Seshat ratio >= 0.80 AND D-PLACE clause holds; KILL = Seshat ratio < 0.60 OR supra_local_share > 0.50; anything else UNRESOLVED. The ANY-FACT rule is the permissive direction for both centralized and hereditary — a polity that at any point within its pre-1500 fact window coded positive counts as coding it; this was chosen for determinism over sparse dating and cuts in the conjecture's favor, which makes a kill that fires DESPITE it more credible, and a confirmation slightly weaker — the adjudicator must weigh it that way, not relitigate it. The hereditary proxy likely biases TOWARD the conjecture: elite status can be hereditary where ruler succession is elective (Mamluk-type and elective-monarchy-type cases would misclassify toward hereditary), so a confirm via the proxy is weaker evidence than a direct succession variable would provide; carry this note into any CONFIRM verdict. 'polity' as a Degree-of-centralization value is uninterpretable and bound as UNKNOWN. Negative Start years are BCE; the Start < 1500 rule therefore admits all BCE polities. Seshat repeats polities across NGA rows — the unit is the unique PolID, once. The count of population-qualified polities excluded for UNKNOWN coding must appear in the resolution ledger next to the ratio, whatever the verdict.
Known priors disclosure: Shepherd-adjacent (2026-07-12): direction anticipated by Gerring et al. 2021 CPS 'Why Monarchy?' and Turchin et al. 2018 PNAS; the sharp two-dataset join (EA033 ethnographic majority-without vs Seshat demographic near-unanimity-with, at a 100k population threshold) is not located as posed. SEQUENCING DISCLOSURE: the operator computed the D-PLACE EA033 marginal distribution before this registration while checking data availability; those numbers were withheld from the registrar, whose thresholds were authored independently beforehand and who bound the code-2 decision blind (see D-PLACE SEQUENCING GUARD). The Seshat side was computed by no one before registration.
Registration community-registration-1-20260712 (bound blind by Fable before computation; docs/generated/community_registration_1_20260712.json) applied mechanically: unit = unique PolID; pre-1500 = Polities Start < 1500; uniform fact-window rule; simple-facts-primary with complex-fallback-must-agree; peak population = max over Value.From/Value.To numerics; centralized = {unitary, unitary state} only; hereditary = {present, inferred present} on the elite-status proxy; ANY-FACT permissive rule; denominator excludes polities UNKNOWN on either variable (32 excluded, reported). D-PLACE: supra-local = codes 2-5 (code 2 bound blind as counting), one code per soc_id, exact fractions, strict inequalities.
Dataset: Seshat Global History Databank, Equinox release (seshatdb/Equinox_Data, June 9 2022 xlsx): 327 pre-1500 polities, 135 population-qualified (peak polity population > 100,000 strictly), joined on PolID; variables 'Degree of centralization' and 'elite status is hereditary' (bound proxy). D-PLACE Ethnographic Atlas EA033 'Jurisdictional hierarchy beyond local community': 1,155 coded societies, fixed CSV, zero conflicting duplicate codes.
computed 2026-07-12