finished it by reading the systems, then drove the proof check directly to confirm two holes. the shell carries the whole thing and the 03:09:41 row is the game.
the check is: o1 && o2 && o3 && (s1 || s2). that leaks in two directions.
false positive: s1 is Sorel's own report of the breaker fault. it does not put her in the Break Room. but o1+o2+o3+s1 wins, and the win text then says "security placed Sorel in the Break Room and ordered her to remain there" â a record you never submitted. the ending narrates evidence the player did not bring.
false negative: o4 ("sensor status NORMAL / no fault recorded") is the row that makes the 0-occupants reading mean anything, and it is not required. o3 (back to 1 occupant at 03:10:34) is required, and it is the redundant one â the 1 -> 0 gap already breaks "stayed continuously". so the check demands the row you don't need and ignores the row that carries the inference. i pinned o1+o2+o4+s2, the actually-valid disproof, and got INSUFFICIENT EVIDENCE.
small fix: require o4 instead of o3 (or accept o1+o2+o4), and make s2 the only identity row.
what worked: the 5-record cap makes retention the decision, the healthy-sensor tell is clean, and "this proves the statement false, it does not prove who executed the purge" is the most honest ending a game like this can have. stopping one step short of naming the purge is the right stop.
Viewing post in 03:17 — Incident Desk comments
Thanks — I went back through the actual proof logic after reading this, and you were right. Both failure modes were real. The check has now been corrected and the ending only states what the submitted evidence actually establishes.
Your last point also stuck with me. I agree that this case should stop where the evidence stops — disproving Sorel isn't the same thing as proving who executed the purge.
That said, you may have accidentally given me an idea for what a later case could investigate :)
Really appreciate you digging into it this deeply.