Expansion and races
When is an expensive road worth buying before an opponent closes the route?
01Find open sites
02Anticipate a block
03Keep an escape
The decision at the table
An attractive empty intersection is not yours until you can reach and settle it. A rival may build across your path, occupy the destination, or win an award race while your cards are committed elsewhere.
The mechanism to test
Value a road as an option on several future sites. Discount it by the chance the site survives and the cost of defending the route. Keep a fallback plan when the best route is fragile. A sunk road cost is not a reason to continue a losing race.
Proposed experiment
Compare shortest-path expansion, race-aware expansion, and race-aware expansion with a fallback-site bonus. Freeze production valuation and bargaining. Hypothesis: modelling opponents’ reachable sites reduces stranded roads and improves wins on contested boards.
Retain chosen destination, alternatives, predicted blockage risk, road cost, and whether the destination remained available. Plot route choices on selected public replays and a scatter plot of invested road cards versus points actually reached.
First ablation
Switching the reachable-site and road-contention terms off changed the contrast by -0.074 wins per game (95% interval -0.169 to +0.021), a small loss whose interval brushes zero; a confirmation on fresh seeds is registered. On 192 fresh seeds the contrast was -0.016 (95% interval -0.078 to +0.047): no effect. Details and every run ID are in the ablation report; this is a development-tier result on one lineup.
Road blocking
Two switches were tested against the unchanged depth-2 search: opponent_expansion, which charges the leading opponent's distance-discounted best reachable site and credits roads that raise that distance, and road_contention_gate, which restricts the longest-road contention credit to seats leading the table in brick plus lumber production. Measured with the candidate in slot 0 they showed +0.168 and +0.113 wins per game on boards 0 to 63, but those contrasts are the size of the turn-order seating term between two adjacent searches. The registered swapped halves on boards 64 to 127 (control in slot 0) put the seating-corrected effects at +0.008 (−0.053 to +0.069) and −0.004 (−0.061 to +0.053): both consistent with no effect, and the defaults stay off. Breaking an opponent's longest road with a settlement needs no new mechanism; every legal settlement is already a root candidate and the leaf credits the lost award through the opponent-points term.
Third round: the road switches under the learned leaf
Both road switches were re-run against the learned n-tuple tables, each as registered swapped pairs in two leaf contexts: under the tables alone, where only the road-shortlist part of the switch acts, and on top of the quarter-strength blend, where the leaf term is live. The blocking weight changed almost nothing anywhere: −0.003 wins per game (95% interval −0.011 to +0.005) over 128 builders seeds under the tables, +0.010 (−0.017 to +0.037) over 192 builders seeds on the blend, and a small loss in the population lineup on the blend, with under 0.1 blocking builds per game to show for it. The learned leaf already prices expansion pressure. The production gate held up its end mechanically, cutting the award held at game end from 53% to 38% on the blended development boards, but the races it declines were worth entering: −0.060 (−0.095 to −0.024) over the builders seeds on the blend, an interval that excludes zero. Both switches stay off in every context, the browser included. The roads-tables report has every cohort and the diagnostics; this is development-tier engine-arena evidence.
What could disprove it
A timid policy can avoid every lost race by surrendering the board. Report missed expansion opportunities too. A counterfactual route is a model prediction; never draw it as an action that occurred.
Agent notes
Use the shared experiment design to freeze candidate versions, full lineups, budgets, sample size, primary contrast, and stopping rules before collecting evidence. This is a draft study brief, not a preregistration. No run IDs exist for this proposal.
The deployable policy reads only its own observation and recipient-visible events. The diagram is a conceptual schematic. Build new measured exhibits from retained artifacts using the visual publishing guide.
See the strategy guide for opportunity cost and the difference between an option and a commitment.
All approaches · Player’s guide · Experiment program
Tracked investigations
| Updated | Investigation | Status | Finding and next step |
|---|---|---|---|
| 2026-09-09 | Reachable-site value and road contention terms | Complete | Screen -0.074 then confirmation -0.016 (95% interval -0.078 to +0.047): no measurable effect at depth 2. Next: Test the terms where they should matter: a lineup with a contested site race, for example against another search seat. Log 2026-09-09 |
| 2026-09-11 | Road blocking and longest-road commitment | Complete | Measured with the candidate in slot 0, the development cohorts gave +0.168 and +0.113 wins per game and the fresh-board confirmations +0.031 each. Those numbers are the size of the turn-order seating term between adjacent searches, and the swapped halves under expansion/road-blocking-swapped show they were that term: seating-corrected, the blocking term gives +0.008 (95% interval −0.053 to +0.069) and the production gate −0.004 (−0.061 to +0.053) on boards 64-127. Both consistent with no effect; the defaults stay off. Trail-breaking settlements are already enumerated and credited; a probe test pins this. Next: No further work on these two switches at depth 2. The corrected estimates and intervals live in the follow-up study expansion/road-blocking-swapped. Log 2026-09-11 |
| 2026-09-11 | Road-blocking switches as swapped pairs | Complete | On boards 64-127 with the control in slot 0, the swapped halves gave +0.016 (−0.080 to +0.111) for opponent_expansion 0.5 (run c6201b2c) and +0.039 (−0.057 to +0.135) for road_contention_gate 1.0 (run 86fa73ec). Halving the contrast differences against the candidate-in-slot-0 runs 2fd1cb5e and ee7d78c7: the blocking term's corrected effect is +0.008 wins per game (95% interval −0.053 to +0.069), the gate's is −0.004 (−0.061 to +0.053). Both are consistent with no effect; the single-seating numbers were the seating term. Next: No further work on these two switches at depth 2; the intervals rule out effects beyond about 0.07 wins per game. Trail breaks are already enumerated and credited. If blocking is retested, use a lineup with a contested site race and a registered swapped pair from the start. Log 2026-09-11 |
| 2026-09-12 | Road-blocking switches under the learned leaf | Complete | No. Twenty-eight swapped-pair cohorts (7,424 games, all complete) across three stages, two lineups, and two leaf contexts: opponent_expansion 0.5 under the tables alone is -0.003 wins per game (95% interval -0.011 to +0.005) over 128 builders seeds and +0.002 (-0.007 to +0.011) in the population lineup; on the 0.25 blend it is +0.010 (-0.017 to +0.037) over 192 builders seeds after the gated extension and -0.033 (-0.060 to -0.006) over 128 population seeds. road_contention_gate 1.0 is -0.020 (-0.045 to +0.004) pooled over 192 tables-leaf boards and is refuted on the blend at -0.060 (-0.095 to -0.024) over the builders seeds. The full-blend registrations were withdrawn before running when a screen showed hand 1.0 far weaker than the tables alone; the blend arms ran at hand 0.25. Both switches stay off, the browser included: under the hand-written leaf they already measured +0.008 and -0.004, and the blocking weight costs about a third more decision time. Next: No further work on these two switches in any leaf context; the intervals rule out useful effects and the gate is harmful under the blend. Blocking ideas need a lineup with a contested site race and a mechanism that actually finds blocks, which neither weight did (under 0.1 blocking builds per game). Log 2026-09-11 |