Find the killers

agrimm2

New member
Joined
Jan 19, 2021
Messages
8
Hello

This is a problem from an Among Us game.

Suppose there are 5 people in a room, you included. You know that 2 of them are killers and you have to eliminate them both. You can only eliminate sequentially. You have 3 people who are suspect( the 2 killers and 1 innocent). If you eliminate the innocent one, you lose, as the killers kill you. If you eliminate one of the killers, he dies and you get to shoot again, with now there being only one innocent and one killer among your suspects.

Lets say you are called Al, and the 4 others are Bob, Carl, Dan and Evan.

The 3 suspects are Bob, Carl and Dan.

To you, each of the suspects has an equal chance of being a killer, BUT you know that between Carl and Dan, at least one of them is definitely a killer.

How do go about deciding who to shoot ?
 
Exactly, How do YOU go about deciding who to shoot ?
This is a free math help forum where we help students solve their problems. This is not a math homework service site where we do homework for students.
If you had read the posting guidelines you would have received help by now since you would have known to share your work with us. If we see your work then we know what method you want to use and would see any mistakes that you made. Armed with that knowledge we could help you in a professional manner.

Here is a hint:

I would reword the problem as follows: There are two killers amongst the three suspects Bob, Carl and Dan. Between Carl and Dan, at least one of them is a killer. You get to shoot twice, if you shoot both killers you live, else you die. How should you go about deciding who to shoot? (we need to assume that if you shoot at someone then they will die)

You can shoot B and C, B and D or C and D

I would calculate the probability of killing the correct pair for all 3 pairs. The one with the highest probability will be the best strategy
 
Thank you for your reply.

My initial thinking was that each of B,C and D had equal probabilities of being a killer, (1/3). I also thought that by knowing that there definitely is a killer among C and D would turn this problem into a sort of Monty Hall problem, but after my initial post I thought about it more and came to the conclusion that it is not the case, as knowing that either C or D is definitely a killer adds no information what so ever. ( My thinking is that if there are 2 killers in a group of 3, then all possible pairs must contain at least 1 killer.)

So now I am stuck again, and the idea that I currently have is that all individuals have equal probabilities. This is what I was hoping to get help on, what am I missing here ?

I maybe should have mentioned that shooting the suspects happens in rounds. So I think that in the first round, you have a chance of shooting a killer of (2/3), and then there would be only one killer left ( and one innocent), so then the probability of shooting the second killer becomes (1/2), and the total probability is (2/6).

I think my initial confusion was thinking there might be an advantage to shooting either C or D first, but now I don't think so anymore.

Also, this is not a homework problem, it was a situation in an actual game and I thought maybe I could figure out if there was an advantage to shooting a particular person first.
 
You are correct that you could have figured out on your own (and did figure out on your own) that amongst C or D, at least one of them was a killer. As you said that does not add any additional information.

It looks to me that you have a list of three names and want to pick the 2 winners (ie killers). Why not pick the complement? That is pick the person not to shoot out (the non killer). What is that probability? Since everything is equal, picking the non killer has a probability of 1/3. So picking the two killers will also be the same 1/3. That is, as you stated, you have a 1/3 probability of randomly picking the two killers. The question is if you can do better than randomly picking the killer--I do not think so.
 
Sim shows that any strategy involving selecting (or not) from the sublist of suspects is no better than picking randomly.
 
Because this comes from a game, I wonder if we have all the information required. It would be odd for a game not to have a correct answer.

Moreover Evan seems to have dropped mysteriously out of consideration. I'd really like to have the complete and exact instructions. There may be a clue that is missing from the problem's paraphrase.
 
Because this comes from a game, I wonder if we have all the information required. It would be odd for a game not to have a correct answer.

Moreover Evan seems to have dropped mysteriously out of consideration. I'd really like to have the complete and exact instructions. There may be a clue that is missing from the problem's paraphrase.

Hey

This scenario comes from a game called Among Us that I watched some people play. In the game there are 10 players, 8 of which are Crewmates ( good guys) and 2 Impostors ( bad guys). The game takes place in a spaceship with different rooms and areas and the Crewmates have to complete a certain number of tasks around the spaceship to win. The Impostors have to kill the Crewmates so that they may win. During the rounds, only Impostors can kill and can only kill Crewmates. Impostors know who the other Impostor is but Crewmates do not. If a player (either Crew or Impostor) finds a dead body around the spaceship, they can report it and the game stops and everyone goes into a meeting where they try to figure out who might have killed the player. At these meetings, a vote takes place to see who gets eliminated. Anybody can be eliminated, but only one person per meeting. If the majority votes for one player, that player is eliminated. If there are ties or no majority for a player, no one gets eliminated and the next round starts. Again, Crewmates win if they manage to complete all given tasks or if they manage to sus out and vote out both Impostors. Impostors win if they manage to kill or vote out enough Crewmates until there are an equal amount of Impostors as Crewmates.

Ok so I was watching a game where there were 7 people were left, 5 Crew and 2 Impostors. In the round, 2 people get killed, a body gets found and all remaining players go into a meeting. Now here is where the problem from my initial post comes from:

From player Al's point of view, Al knows he is Crew, and Al had spent all of that round with Evan, so he is absolutely sure Evan did not kill anyone that round. He also know that both kills are done by different people because the game has a setting where after an Impostor kills, there is a "cooldown" period where he cannot kill again for 25 seconds, and that round lasted less that 25 seconds. So it could not be the case that one Impostor kills, cools down and kills again, so the 2 kills must have been done by 2 different people. Since Al knows he is Crew, and now knows Evan is also Crew, this leaves the 2 killers to be between the rest of the people (Bob, Carl and Dan). By a few details of this particular game, Al and Evan deduce that one of the kills must have been done in an area where Carl and Dan were ( This is why in my initial post, I said that it is known that between Carl and Dan, one is definitely the killer).

As was shown in previous posts in this thread, this was not useful information. I created the first post in this thread because I wanted to see, given this scenario, if there was a strategy for Al where he could sus out the impostors with higher probability, but now I don't think so.

In the vast majority of Among Us games, concepts like equal probabilities, randomness and independence do not apply, because players decide who to vote for in meetings based on influence from other players and previous rounds and personal suspicions, but in this one particular game, the player Al really did not lean one way or another among the 3 suspects, and so assigned equal probabilities to all three suspects.

This is a long post, I don't know it makes things clearer for you, but I hope it helps.

Thanks to everyone else who engaged.
 
Because the game has multiple rounds and the probabilities are changing each round, Bayes Theorem might be useful in principle. Whether anyone has time or energy to do those computations in between beers is a different question.
 
Top