ADR-0005 — Emergent doom waves: author causes, never outcomes; seed = RNG + schedule

Status: ACCEPTED · 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

Doom arrives in waves. The fork: authored waves (designed timings — pacing guaranteed, sim partly theater) vs emergent waves (arising from opponent behavior plus the player's accumulated ledger — pure, attributable, pacing at the mercy of the dynamics). Pip chose emergent, with a thumb on the scale: "I think the doom's waves are emergent. We might lean our thumbs on the scale at certain points… For example, rival_lab could get a wave of funding at point X."

Decision

Author causes, never outcomes

The designer's thumb may touch inputs to the sim — a funding wave hits rival_lab at turn 9, a regulation lapses, a key researcher defects — and the wave that reaches the player is whatever the sim does with that cause given opponent state and the player's ledger. The thumb never touches the doom variable directly. This one rule keeps three prior commitments intact:

A seed = RNG seed + event schedule

This redefinition is also the cheap implementation of seasons as scenarios: a "compute-overhang week" is a schedule dense with compute-cost causes — no engine changes. Patches rotate the meta by rotating world-model assumptions, which is more honest than nerfing numbers. Feedstock: the pdoom-data repo (1,194 curated real AI-timeline events, actively maintained) — real events, lightly fictionalized, as schedule entries. This closes Pip's "historical events affecting the game" patch ambition with existing infrastructure.

Seed vetting replaces wave authoring

The cost of emergent, stated honestly: pacing guarantees weaken — the same cause produces different waves depending on state, so some seeds will be degenerate (dead by turn 3, or a snoozefest). Curation replaces authorship: run candidate seeds through baseline_simulator.gd headless (no-action baseline plus crude heuristic bots) and reject seeds whose doom trajectory falls outside a playability envelope before publishing. Pip is a seed curator, not a wave author — the SimCity thumb with quality control, built once.

Rejected: player-triggered disasters

Pip, explicitly: hardness is baked into game settings for the engine, never a player-facing disaster menu. (Player-chosen difficulty modifiers at run start are a separate parked item — ADR-0008.)

Beacons