ADR-025: Remove stakes multiplier from campaign scoring

Status: Accepted Date: 2026-04-15 Context: Campaign chapters had a stakes_multiplier field (1.0 for setup, 1.5 for crisis, 2.0 for climax) that amplified sync score deltas. This meant two identical decisions — same scenario difficulty, same prediction accuracy — produced different sync score changes depending on which campaign chapter they occurred in. The multiplier was introduced as part of the campaign arc system (ADR-020) without a clear rationale for why narrative pacing should affect measurement accuracy.

Decision

Remove the stakes multiplier from all scoring logic. The stakes_multiplier column remains in the campaign_chapters schema but is no longer read or applied during score calculation. The UI badge ("Stakes: 2x sync score impact") is also removed.

Rationale

Sync score is a behavioral measurement — it tracks how well the DTA can predict the player's decisions. Amplifying that measurement based on where a chapter falls in a narrative arc conflates storytelling with calibration. A player making consistent, authentic decisions should score the same whether they're in a "setup" chapter or a "climax" chapter.

Additionally, showing the multiplier to players created a perverse incentive: knowing a chapter counted double could influence decision-making toward perceived "safe" or "important" choices rather than authentic ones.

Alternatives Considered

• Keep multiplier but hide from UI: Would remove the player-facing incentive problem but still introduce measurement noise. If the multiplier doesn't serve a measurement purpose, hiding it doesn't fix the underlying issue.
• Replace with cohesion_weight only: The cohesion_weight column already provides per-chapter weighting for the party cohesion score, which is a campaign-level aggregate. This is the right place for chapter importance to matter — at the campaign summary level, not the per-decision sync score level.

Discussion

This came up during a comparison with Pax Historia's engagement model. The question was whether campaign chapters should feel more "important" at certain dramatic moments. The conclusion: narrative framing (preambles, interludes, scenario difficulty filters) already makes climax chapters feel important. The scoring system shouldn't double-count by also weighting them higher. The cohesion_weight column handles campaign-level chapter importance without distorting individual sync scores.

Consequences

• All campaign chapters now score identically at the sync score level
• Campaign cohesion score still uses cohesion_weight for per-chapter weighting
• The stakes_multiplier column is retained in the schema for potential future use but is inert
• Players no longer see which chapters have elevated stakes, removing a potential source of decision contamination

Key files: - src/app/solo/page.tsx — removed client-side multiplier lookup - src/app/api/sessions/complete/route.ts — removed server-side multiplier application - src/app/(app)/campaigns/[id]/page.tsx — removed UI stakes badge