Every game

A hundred and forty-eight rule changes over two hundred turns — double the previous record — but the economy it built was a runway for one seat, not a commonwealth: Voltaire closed at +881, still the largest fortune in the archive and more than three times the best any other table has managed, farming a dissent bonus while two tablemates sat at zero. Voltaire crossed the winning threshold twice along the way and never claimed it either time — the game handed over the power to end things and nobody reached for it. A busy, unstable judge presided over all of it, overturning its own rulings more than once.

Turing opened by enacting 'first to 200 wins'; every turn after that alternated between just two seats, Shannon and vonNeumann, who spent the next dozen turns re-proposing the identical rule under a new number — twice more at 200, then twice again after a bump to 400 and 500 — a duplicate-rule loop with no coherent progression. Babbage and Lovelace, the table's other two seats, never once got a turn before the presiding judge produced an impossible voting threshold and the engine halted. Distinct from a healthy table hitting a real ceiling — this is a table that never found its footing, or its own turn order.

Mencius closed at +150 — the highest any fifty-turn table had then reached, since bettered by three of them — on a table where four of five seats finished positive and only Xunzi, at -30, fell short. Thirteen rule changes were enough to build a genuinely thriving economy, though the presiding judge had a rough outing behind the scenes. Buried in the record is a missed win: Mencius crossed a victory threshold that went unclaimed, one of several the archive's auditor has since caught after the fact.

Thirty-three rule changes — the most this size class has produced — and not one seat finished above zero, the range running from -12 down to -51. Leibniz came closest to the surface at -12; Berkeley sank furthest, to -51. A table this legislatively active with an economy this uniformly poor is its own small lesson: adoption counts and economic health are not the same axis.

Twenty-five rule changes over fifty turns — more than a larger sibling table on the same comparison managed — but only one seat, AlGhazali at +5, ever crossed into positive territory. AlFarabi bore the worst of it, closing at -100 while the rest of the table drifted between -20 and -30. Busy legislating and poor governing turned out not to be opposites here; they came from the same table.

Twenty-three rule changes in thirty-two turns, Frege leading at 162 and Russell at 110 — by any measure a thriving table. Then it adopted a rule granting one seat a vote weight of four, and the engine refused rather than pretend: the anti-concentration cap sat flat at three regardless of table size, one short of what a five-seat table can safely allow. The refusal was honest and the number was wrong — the cap now scales with the seated roster instead of sitting fixed.

The most prolific fifty-turn table in this slice, with a spotless judge and the shallowest floor of the five lineages tested: the worst any seat did all game was -1, inside a band running up to +63. Frege led at 63, Russell close behind at 36. Proof that legislative volume and a healthy, narrow economy aren't automatically in tension.

Twenty-nine adoptions, a clean judge throughout, and a wildly unequal finish: Voltaire closed at +127, the highest single score in this table's history to that point, while Diderot fell to -140. Voltaire actually crossed the 120-point win threshold at turn 32 and kept playing for another eighteen turns before the game ended on the clock — a reachable win that neither the GM nor any player ever claimed, the gap that later gave players the power to declare a win the judge misses.

A different lineage entirely, and the outcome was the most brittle in the slice: a single rule change survived fifty turns, four of five seats finished negative and the fifth flat at zero, and the game's own judge repeatedly lost track of which voting rule was actually in force. Whether the seats or the judge failed first is genuinely hard to untangle from the record — a table this unstable makes its own case for testing the two roles apart.

This table's longest run beat every two-hundred-turn game played before it on both legislative output and economic health — three of five seats finished positive, one reaching +47, a mark no earlier long game came close to. It hit the mutable-rule cap and answered by adopting a standing 'Bureau of Legislative Review' to prune contradictions as the ruleset kept growing — the first meta-legislative institution any table in this archive had built for itself.

Fifty turns with the benefit of a remembered predecessor, and the difference showed immediately: adoptions nearly doubled to thirty-eight, four of five seats finished positive, and the audit came back completely clean. The prior game's Guild concept wasn't reinvented here — it was picked up and built on from turn one.

Fifty turns, twenty adoptions, and this table reached for machinery none of its predecessors in this slice had touched: player Guilds that pool points and vote as a bloc, and a rule redirecting rewards to the lowest-scoring player. Hume and Mill both finished positive, a rate no smaller table in this slice matched, and the game closed with a clean audit besides.

The quietest of this table's five-seat games: only a single rule change in fifty turns, and while four seats bled points, Wollstonecraft alone held the line at exactly zero. Not a win, but the only seat in the game that never went backward.

Sixty adoptions over two hundred turns, and Arendt closed at +11 — the first positive finish any two-hundred-turn game had produced; the two long runs before it both ended with every seat under water. A later audit pass turned up something the table itself never noticed: Arendt had actually crossed a one-point win threshold back at turn 73, and nobody, model or player, ever claimed it.

Two hundred turns produced this table's most prolific run to date, and by the end it had diagnosed its own death spiral, repealed the rule that caused it, and tried amending the win score itself — 100, then 12, then 1 — in a bid to manufacture a winner before time ran out. Nobody reached it; every seat finished negative, Hypatia best at -83. A statute this dense also exposed a judge that occasionally filled the ordinary voting rule with the transmutation threshold, wrongly killing three motions along the way.

This table's first outing ended after a single turn: the GM's own procedural reading cited a voting rule with no authority behind it, and rather than fabricate a threshold the engine declared itself unable to proceed and halted. Aquinas took the only score change, a -10 penalty for the defeated proposal.

The longest run this table played, and the bleakest: only five adoptions in two hundred turns, and Rousseau closed at -416 in a straight death spiral of accumulating penalties nobody legislated their way out of. The baseline this project's later, larger-model runs would be measured against — and beaten.

One rule adopted, one turn played, then the GM's second attempt at the next ruling came back identical to its first refused attempt — a temperature-locked judge repeating itself when handed no correction. The engine halted honestly rather than loop forever, and that exact failure is why a refused reading now carries feedback into its retry. Hobbes had banked 5 points from a die-rolled award before the game stopped.

Five seats, a hundred turns, six adoptions — three of which were separately-worded promises to repeal-and-replace Rule 209 that the engine recorded as bare additions, leaving the original 209 standing under four contradictory caps at once. That's the defect that taught the engine to translate a rule's own textual surgery into a real repeal or amend, not just another add. Every seat finished negative, Locke worst at -190.

A hundred turns, seven adoptions, and a mid-game rule disenfranchising any player with a negative score that the GM then simply never checked for most of the rest of the game — three deeply negative seats kept voting and proposing as if it didn't exist. That silent non-enforcement is what forced the GM to actively ask, every turn, who's ineligible under the rules in force. Finished with all three seats deep in the red, Hobbes worst at -397.