ADR-0007 — Alliances: the third client of Ledger + SA (treaty = shared liability + shared sight)

Status: ACCEPTED as design; build third, strictly after ADR-0003 and ADR-0004 · Decided: 2026-07-04
These are the actual decision records the game is built from, published unedited except for internal process notes. They describe why the game works the way it does. They are not a strategy guide, and some of them argue with each other.

Context — why the original deferral was wrong in kind

The pre-workshop triage parked alliances/voting as a "big isolated system." Pip's own purpose statement convicts that: "there isn't really consensus anywhere on the best approaches on how to deal with AI risk, other than try to coordinate pause / stop functions." Coordination is the one lever the real-world model endorses — and the game's strategy space omitted it. For an artifact that teaches the way Civ teaches, a game with no coordination mechanic quietly argues coordination doesn't exist. That is a structural falsehood, not a missing feature. What was right in the deferral was sequencing: alliances need almost no new machinery because they are made of the other two systems' parts.

Decision

A treaty is a shared liability plus shared sight: mutual ledger entries (obligations with fuses; defection = the exposure event running in reverse) plus a mutual SA channel (alliance-sight: slow, public, obligates you).

Votes are scheduled causes (ADR-0005): a treaty vote at turn X is visible terrain, with more lead time if you've bought it (ADR-0004).

The whip loop — from Pip's House of Cards image: "there are critical votes coming up, and people are wavering. have you got them?" Whipping a vote means spending money, reputation, and favors to acquire pledges — receivables in the two-sided ledger (ADR-0003). A pledge is not a vote until cast: receivables carry counterparty risk (a lab under financial pressure wavers; a rival counter-whips; a scandal flips a signatory). Wavering is computed from counterparty state, never scripted — author causes, never outcomes: the vote result is simmed. "Have you got them?" is a sight question about your own receivables, which gives alliance-sight a job only it can do and passes the decision-flip test trivially (knowing who wavers changes whether you shore up, buy a replacement vote, or write the treaty off).

Influence is not a resource (decided by Pip): no global influence stat — a fourth currency would be the parallel-economy trap, and less thematic (leverage is over this lab, that actor). Influence is the receivables column, per-actor. Reputation stays global and public. Fallback on record if playtests show per-actor favors overwhelm players: a global influence pool as the legibility compromise — build the honest version first.

Sequencing discipline (binding)

Built third, after the Ledger and outward SA, because it is assembled from their parts. The vote screen must not be built early because it is dramatic — a vote UI without a working ledger underneath is theater.

Beacons