Settlers / Research

73 pages · Search titles and descriptions

↑ ↓ to navigate · Enter to open · Esc to closeLocal search
Play the game

The learned leaf through the protocol arena

Win rate by competitor

95% interval
ntuple-leaf · slot 057.5% · n=80
expectimax-v2-plan · slot 131.3% · n=80
eta · slot 26.3% · n=80
fast · slot 35.0% · n=80
0%50%100%
Run e861616e-8100-43e6-a8d3-39047b5b6669. 80 completed of 80 planned games. 95% Wilson intervals; descriptive evidence only.

What was registered

Experiment a5e9b3f1-8ace-418f-b8e8-bd54c825c5c9 was registered under patterns/ntuple-leaf before the candidate was frozen: the frozen tables at artifacts/ntuple/hex-portfolio-main.bin as the leaf of the baseline's exact search configuration (depth 3, 1.5 s budget, 8 scenarios, 4 sampled worlds) in slot 0, expectimax-v2-plan in slot 1, eta and fast in the other seats, on seeds 320 to 339 in deterministic mode, 80 games. The rule: all games complete validly with no server timeout moves; support only if the 95% interval of the paired per-seed win difference between slots 0 and 1 lies above zero.

Run e861616e-8100-43e6-a8d3-39047b5b6669 played all 80 with twelve matches in flight. The two searches ran as separate remote processes against the server and saw only their own seated view and event history; the tables were loaded from the file named in the policy registry, whose SHA-256 (f8e61e66…) is recorded with the candidate's receipt. Games averaged 73 turns.

Slot / policyWinsCompletedWin rate95% Wilson intervalMean points
0 / ntuple-leaf468057.5%46.6% to 67.7%8.28
1 / expectimax-v2-plan258031.2%22.2% to 42.1%6.90
2 / eta5806.2%2.7% to 13.8%6.10
3 / fast4805.0%2.0% to 12.2%6.03

The result agrees with the engine-arena screens: at depth 3 against depth 3 on 64 fresh engine seeds the contrast was +0.242, and here through the protocol on 20 seeds it is +0.263.

One game, viewed from above

The first match of the cohort, seed 320 in its first rotation, which the tables' seat won in 65 turns. It is a real recorded game chosen for the replay; it does not by itself establish which policy is stronger.

ntuple-leaf slot 0’s turn

Turn 0 · Place a settlement · 10 points to win

Ore4Ore, rolls 4, 3 of 36 dice outcomesWool11Wool, rolls 11, 2 of 36 dice outcomesBrick12Brick, rolls 12, 1 of 36 dice outcomesOre8Ore, rolls 8, 5 of 36 dice outcomesBrick3Brick, rolls 3, 2 of 36 dice outcomesBrick6Brick, rolls 6, 5 of 36 dice outcomesOre9Ore, rolls 9, 4 of 36 dice outcomesGrain5Grain, rolls 5, 4 of 36 dice outcomesLumber10Lumber, rolls 10, 3 of 36 dice outcomesWool11Wool, rolls 11, 2 of 36 dice outcomesWool5Wool, rolls 5, 4 of 36 dice outcomesDesertDesert, blocked by the robberLumber2Lumber, rolls 2, 1 of 36 dice outcomesWool9Wool, rolls 9, 4 of 36 dice outcomesLumber4Lumber, rolls 4, 3 of 36 dice outcomesGrain10Grain, rolls 10, 3 of 36 dice outcomesGrain6Grain, rolls 6, 5 of 36 dice outcomesLumber3Lumber, rolls 3, 2 of 36 dice outcomesGrain8Grain, rolls 8, 5 of 36 dice outcomes3:1Any resource 3:1 port2:1Lumber 2:1 port2:1Brick 2:1 port2:1Ore 2:1 port2:1Grain 2:1 port3:1Any resource 3:1 port3:1Any resource 3:1 port3:1Any resource 3:1 port2:1Wool 2:1 port
Bank19Brick19Lumber19Wool19Grain19Ore25 development cards left

ntuple-leaf slot 0

Playing

0 VP
0 resource cards0 development cards0 settlements0 cities0 roads0 knights played
Longest route: 0

expectimax-v2-plan slot 1

0 VP
0 resource cards0 development cards0 settlements0 cities0 roads0 knights played
Longest route: 0

eta slot 2

0 VP
0 resource cards0 development cards0 settlements0 cities0 roads0 knights played
Longest route: 0

fast slot 3

0 VP
0 resource cards0 development cards0 settlements0 cities0 roads0 knights played
Longest route: 0

Actions & conversation

1 events · since the previous position

  1. Table

    The game started.

Game 1 in the scheduled cohort.

120 of 164 recorded positions. Board changes between positions are grouped.

Replay sources

Game 9d5618a5-c94e-4311-a2a4-9ad40d0b1b12 · version 14

Frames: 20d74cd3af2c892ed69591e1c0479164c3bb76762bddc7ca15ac8a519955902c

Public events: 960d87fb6e1267996c409829d57731e18223ae26db392ee53887fc9b34faeb5b

Limitations

One lineup, one seat rotation scheme, and twenty seeds. The interval is wide because the cohort is small; the engine screens on more seeds carry the sharper estimate. Both searches negotiate only with the fixed builders' willingness model, so nothing here speaks to trade-rich tables with negotiating opponents or to play against humans. The tables decide more slowly than the hand-written leaf under the same budget, which the 1.5 s cap absorbs at depth 3 but which would matter under a tighter deadline.

Run record · Public frame source SHA-256: 20d74cd3af2c892ed69591e1c0479164c3bb76762bddc7ca15ac8a519955902c.