Suppose you and 6 your friends got caught by some mad scientist. He tells you that he will put all of you into Trolley Problem, but as proper scientist should, he would assign all the 7 roles at random. You have some time to talk through your strategy. What should you commit to?
Like, the obvious and fair answer is to commit, whoever ended at the lever should pull it unconditionally. Even if mad scientist secretly picked the one on the lever and the one to die already, it’s a conclusion you should end up with.
The Problem with this line of thinking in more realistic situations is that it’s 1 easy to fudge the results without actual pre planning and mutual policing 2 the losers might have a priori larger chance to be losers and winners winners, so executing such tradeoff without explicit negotiation would often result in costs being passed on people who would have not agreed on such deal.
Like, suppose the scientist said, “You, in the brown shirt! You in particular would have 50:50 chance of being put on solo track”—then, what would be the fair agreement?
I guess you still can fallback on updateless thing, but then you would have to model what happens if the scientist names other people, would people agree on following the equal split? and how likely the scientist is to name different people? etc. It’s kinda unclear how to deal with this.
Why on earth should we believe what the mad scientist says? In the proposed scenario, he has all the power; he can tell us anything and then do something different. We should each commit to doing whatever seems like it will free the most of us to get revenge upon him, so that any future mad scientists will not abduct people for their mad experiments.
That’s cheating, man! It’s fighting the hypothetical.
You can postulate that there is idk, mad science institute that the scientist submits proposals to and they execute his experiments impartially. Or there is magical truth spell. Or some other commitment mechanism.
Or you just found yourself trapped in a mechanism that has this property, that situation coming to exists accidently.
The mad scientist here is just a tool to make the clean self contained example.
But that’s the thing: it doesn’t matter if the mad scientist is telling the truth; your strategy is still to maximize the number of survivors capable of seeking justice.
See, you think the mad scientist is putting you in a game-theory thought-experiment with a weird symmetry-breakage. I think the mad scientist is putting himself in a game-theory thought-experiment with a weird symmetry-breakage. And people shouldn’t go around doing that!
That’s fine, I guess, but besides the thing that of interest to me. Like, suppose there is no mad scientists, you just randomly stumbled on piece of industrial machinery, that has this property. E.g. it will give random one of you ability to divert it from 5 others to kill another.
Then, in another variant, one of them is at more risk to be singled out, etc.
Suppose you and 6 your friends got caught by some mad scientist. He tells you that he will put all of you into Trolley Problem, but as proper scientist should, he would assign all the 7 roles at random. You have some time to talk through your strategy. What should you commit to?
Like, the obvious and fair answer is to commit, whoever ended at the lever should pull it unconditionally. Even if mad scientist secretly picked the one on the lever and the one to die already, it’s a conclusion you should end up with.
The Problem with this line of thinking in more realistic situations is that it’s 1 easy to fudge the results without actual pre planning and mutual policing 2 the losers might have a priori larger chance to be losers and winners winners, so executing such tradeoff without explicit negotiation would often result in costs being passed on people who would have not agreed on such deal.
Like, suppose the scientist said, “You, in the brown shirt! You in particular would have 50:50 chance of being put on solo track”—then, what would be the fair agreement?
I guess you still can fallback on updateless thing, but then you would have to model what happens if the scientist names other people, would people agree on following the equal split? and how likely the scientist is to name different people? etc. It’s kinda unclear how to deal with this.
Why on earth should we believe what the mad scientist says? In the proposed scenario, he has all the power; he can tell us anything and then do something different. We should each commit to doing whatever seems like it will free the most of us to get revenge upon him, so that any future mad scientists will not abduct people for their mad experiments.
That’s cheating, man! It’s fighting the hypothetical.
You can postulate that there is idk, mad science institute that the scientist submits proposals to and they execute his experiments impartially. Or there is magical truth spell. Or some other commitment mechanism.
Or you just found yourself trapped in a mechanism that has this property, that situation coming to exists accidently.
The mad scientist here is just a tool to make the clean self contained example.
But that’s the thing: it doesn’t matter if the mad scientist is telling the truth; your strategy is still to maximize the number of survivors capable of seeking justice.
See, you think the mad scientist is putting you in a game-theory thought-experiment with a weird symmetry-breakage. I think the mad scientist is putting himself in a game-theory thought-experiment with a weird symmetry-breakage. And people shouldn’t go around doing that!
That’s fine, I guess, but besides the thing that of interest to me. Like, suppose there is no mad scientists, you just randomly stumbled on piece of industrial machinery, that has this property. E.g. it will give random one of you ability to divert it from 5 others to kill another.
Then, in another variant, one of them is at more risk to be singled out, etc.