The central object here is a board. You collect statements from nine witnesses, pin them up, and connect the ones that cannot both be true. When it works — and from about hour four it works very well — it is one of the most satisfying deduction systems we have used.
The opening explains all of it in the first eighteen minutes, while your board is empty.
Four rules, delivered too soon
| Rule | Explained at | First usable at |
|---|---|---|
| Pinning a statement | 3 min | 5 min |
| Contradiction links | 7 min | ~40 min |
| Corroboration chains | 11 min | ~2h 10m |
| Retraction handling | 15 min | ~3h 25m |
The first rule is fine: taught, then immediately used. The other three describe operations you cannot perform for between half an hour and three and a half hours, and two of them are explained a second time in a later panel, presumably because the developers suspected the first attempt had not landed.
They were right to suspect it. The second explanation does not help either, because it is also delivered before the need.
The retention check
At ten hours we tested all four rules cold. Pinning and contradiction links were intact. Corroboration chains and retraction handling were not — we knew the features existed, could not have described how to use them, and had in fact been playing for six hours without touching either.
The system is genuinely good. We arrived at it late and by accident, having been told about it twice.
Explaining a mechanic twice before it is available is not twice the teaching. It is the same failure repeated at greater length.
What the fix looks like
Almost nothing. The corroboration rule needs to arrive at the moment two witnesses first agree about something, which the game can detect trivially since it constructed the case. The retraction rule needs to arrive the first time a witness changes their story, which is a scripted event.
Both moments already exist in the game. The instruction simply sits an hour and three hours ahead of them respectively, and moving a panel is not an expensive change.
What the opening does well
The atmosphere is immediate and excellent. Within ninety seconds you know what kind of story this is, what your role in it is, and that the rain outside the window is going to continue for the entire game. Tone is taught faster and better here than in anything else in the file.
The first witness interview is also very well judged: short, contradictory, and impossible to resolve, which teaches you to distrust testimony before any rule mentions unreliability. That is a shown lesson in a mostly told opening, and it is the best minute of the eighteen.
The rest of the game
Strong. Twenty-nine hours, nine witnesses, and a case that holds together under scrutiny — we went back through our board at the end and found no cheats. The writing is dry in a way that suits the material, and two of the witnesses are genuinely memorable characters rather than testimony dispensers.
Weaknesses: the middle sags around hour twelve when the case stalls on a single piece of evidence, and the ending explains itself more than it needs to, which is the same instinct that damaged the opening showing up again at the other end.
| Element | Assessment |
|---|---|
| Deduction system | Excellent once reached |
| Rule timing | Three of four far too early |
| Retention at ten hours | 2 of 4 |
| Tone established | Within 90 seconds |
| First interview | Best teaching in the opening |
Where this leaves it
Recommended, with the observation that its best system is hidden behind its worst decision. Nothing here needs redesigning; three panels need moving. It is the most easily corrected failure in the file and the clearest demonstration that ordering matters more than clarity.