ADR-068: Multiplayer scenario quality parity with solo

Status: Accepted (retroactive) Date: 2026-05-23 (decided May 11–12; documented retroactively) Context: A persistent question through May 11–12: should multiplayer scenarios be held to the same quality bar as solo scenarios, or is some quality reduction acceptable for multiplayer-themed or team-context-generated content? ADR-052 (deferred) covered multiplayer generation constraints; ADR-064 (proposed) addresses difficulty. Neither closes the quality-parity question. The conversation audit (item 19) surfaced this as a load-bearing operational principle without ADR backing.

Decision

Multiplayer scenarios are held to the same quality bar as solo scenarios. No exemptions for multiplayer-themed scenarios, team-context-generated scenarios, or any other multiplayer-specific surface.

Specifically: - All ADR-061 pre-commitment planning requirements apply - Scan strips required where solo scenarios require them - Cast blocks required at the same characterNames ≥ 2 threshold - Voice contract (ADR-034) applies identically - Content guardrails (ADR-027) apply identically

If a multiplayer scenario can't meet solo quality, regenerate or omit — do not ship at lower quality.

Rationale

From the May 11 conversation: "technically the scenario for multi should be as high calibre as a single player scenario regardless of theme."

Three reasons the quality bar holds:

1. Multiplayer is a trust-building mode (ADR-066). Low-quality scenarios in a trust mode are worse than low-quality scenarios in a discovery mode — they degrade the very signal that makes multiplayer worth playing.

2. A user playing both solo and multiplayer should not perceive a quality drop. If multiplayer feels worse, users self-select away from the trust-building surface and back to solo, defeating the multiplayer mode's purpose.

3. The corpus is shared. Many scenarios serve both surfaces. A two-tier quality model would require a per-surface quality flag and corpus-side enforcement that doesn't currently exist (and isn't worth building).

Alternatives Considered

• Lower quality bar for multiplayer-themed scenarios specifically: Rejected. "Multiplayer-themed" isn't a useful corpus-side distinction; most scenarios fit any mode.
• Two-tier quality with explicit per-scenario tagging: Rejected. Adds tagging overhead and corpus-side enforcement work for a question that doesn't have evidence-based answers (no playtest signal suggests multiplayer needs different content quality, only different content selection).
• Looser bar in early-game multiplayer for onboarding: Rejected. ADR-064 handles onboarding via the campaign ramp, not via quality reduction.

Discussion

This decision is best understood as a corollary of two existing ADRs: - ADR-066 (multiplayer = trust mode) establishes the stakes - ADR-064 (multiplayer keeps full difficulty; campaign is the onboarding ramp) establishes the principle that multiplayer is the "real game"

If multiplayer is the real game, multiplayer content has to be the real content. The quality-parity decision falls out of those two principles, but stating it explicitly prevents future contributors from optimizing the multiplayer surface for "more scenarios faster" at the cost of quality drift.

Consequences

• Multiplayer scenario generation tooling must use the same validation pipeline as solo
• No "multiplayer-lite" content surface; if a corpus expansion is needed for multiplayer themes, it expands at full quality
• Trigger to revisit: if playtest signal shows users specifically wanting lighter content in multiplayer mode (not just easier — different in kind), revisit. No such signal exists as of 2026-05-23.

Key files: - src/lib/ai/scenario-generator.ts — shared generator used by both surfaces; quality-parity is enforced via shared validators - docs/decisions/064-multiplayer-difficulty-campaign-onboarding.md — companion ADR - docs/decisions/066-multiplayer-mechanics-extending-adr-033.md — companion ADR - docs/research/audits/conversation-decisions-audit.md — source item 19