How to solve a one-liar logic puzzle
In a one-liar puzzle everyone tells the truth except one person, so the fastest way in is a clash: two statements that can't both be true mean the liar is one of those two people, and everyone else is cleared at once. Then test the two remaining suspects one at a time: assume a suspect is honest, and see whether their story still fits everything else.
Updated · The Open Apps, Inc.
A method that works
- Start at the incident. Ask about the minute it happened. Everyone will deny being at the scene, which is expected, but their answers place them, and places can clash.
- Hunt for a clash. “I was alone in the Kitchen” against “I was in the Kitchen with Ada” can't both be true. The liar is one of those two speakers.
- Use the map and the clock. In Alibi Hour a person can only move one room per ten minutes. If an honest answer puts someone two rooms away ten minutes before the incident, they couldn't have been at the scene.
- Assume, then test. Pick one of the two suspects and assume they're honest. If that makes the other person's story impossible, you have your culprit; if it makes everything fit, look again.
- Stop as soon as it's proven. Every extra question costs. Accuse once only one suspect can still be the liar.
Clues and what they prove
The patterns that come up most, and what each one rules out.
| Pattern | What it proves |
|---|---|
| Two statements that can't both be true | The liar is one of the two speakers; the other two suspects are cleared |
| An honest answer puts someone elsewhere at the incident minute | That person isn't the culprit (if the speaker is honest) |
| Someone is placed two rooms from the scene ten minutes before or after | They couldn't reach the scene in time |
| A suspect's own story needs a twenty-minute walk in ten | That suspect is lying |
| Everyone's answers fit together | Ask somewhere else: the clash is at another time |
These are the rules Alibi Hour's solver uses; every case has exactly one culprit under them.
Use it in ChatGPT
Alibi Hour is a ChatGPT plugin. Type one of these:
- “Let's play today's Alibi Hour case”
- “Ask Dev where he was at 7:40”
- “Who could still be lying, given what I've asked?”
Coming soon to ChatGPTHow it works in ChatGPT
Questions
- Does the liar lie in every answer?
- In Alibi Hour, yes: every one of the culprit's statements is false, and every other suspect's is true.
- What if two people say they were together?
- If both are honest, they were. If one is the liar, the claim is false; check what the third person in that room says.
- Is guessing allowed?
- Yes, but a right guess your answers didn't prove counts as lucky, not proven.
- How is par worked out?
- A solver tries every set of questions, smallest first, and finds the fewest that leave only one possible culprit.