Make information public enough to survive
Investigative results, protection feedback and visit logs should reach the town before the source dies. That does not mean every role full-claims immediately; it means valuable evidence should not exist only in one player’s memory.
Build independent confirmation
A Sheriff result is one clue. A Lookout visit plus a role claim plus role-list compatibility is a chain. Town should prefer cases supported by different mechanics because evil can more easily fake one story than several independent interactions.
Organize claims before the trial clock
Use daytime pressure to collect roles and wills. Once a player reaches trial, the lobby should already know what it wants explained. Randomly opening five new claim threads during defense helps evil hide in noise.
Protect the information network
Doctor, Bodyguard, Jailor and control roles should preserve players whose future actions or confirmed votes matter. Protection is not about prestige; it is about keeping the evidence system alive one more day.
Do not hand away parity
Late-game Town should calculate the post-vote and post-night player count. A miselimination at six can become evil parity before Town receives another useful result. When a safe information line exists, preserve the next day.
Town coordination example
Sheriff finds Player 8 suspicious. Instead of instantly voting, Town asks for the full claim. Lookout has a visitor log placing Player 8 at the victim, and the claimed role should not visit. The role list also has no open slot for the claim. Three different sources now support the same elimination.
Keep a claim ledger that preserves uncertainty
| Player | Claim / observation | Next question |
|---|---|---|
| 3 | Sheriff; N2 found 8 suspicious | Who can corroborate the target or explain altered feedback? |
| 8 | Doctor; says N2 healed 11 | Was the will posted before the accusation? |
| 11 | Acknowledges a heal-related event | Does the feedback establish a heal, or only survival? |
This ledger does not convict Player 8. It identifies the next useful questions without confusing a claimed result with a confirmed event. Update the record when answers arrive instead of restarting the argument from memory.
Common mistakes
- Keeping key results private until trial.
- Letting one investigative result substitute for a full case.
- Running multiple unrelated wagons at once.
- Ignoring the next night’s parity after a vote.
