The point would be that model has to be easily falsifiable—you should be able to see if it doesn’t simulate amoeba.
The Pascal’s mugging is not easily falsifiable. I think of myself as a fairly intelligent individual; it took me 4 days to think that—whoah, the number of possible beings is not as huge as 3^^^^^3 ; if a being is made of 10^30 components each of which can be in either of 10^30 states, the number is only (10^30)^(10^30) which seems way smaller (not sure though, the knuth’s up arrow is unwieldy to compare to anything). And that’s some seriously frigging complex being I’m talking of (to human as human is to amoeba). I can’t rely on myself to falsify such stuff reliably.
The issue with pascal’s mugging is that the refusal to simulate amoeba only decreases the probability of validity of the claim by a constant factor, and the mugger can stack more uparrows. Suppose that one in ten thousands superbeings refuses to demonstrate to you the simulated amoeba despite being capable of doing so.
Also, what if it simulated amoeba but claims it’ll torture 3^^^^^3 humans? I can falsify the latter part because there can not be this many distinct human mind states, but only after taking a lot of time to think about the claim.
It seems reasonable to me that my confidence that the mugger can do what it claims to should be inversely proportional to the discrepancy between what the mugger claims and what the mugger is willing to demonstrate. Why do you say it should be a constant factor?
What if one in ten thousands genuine super-beings is not a nice guy and refuses to demonstrate you stuff? The refusal is only linked to ability for those unable to demonstrate, or those barely able to demonstrate. If you mail world’s top mathematicians a question what is 2*3+4 , very few will write you back; you may actually get higher willingness to ‘demonstrate’ ability to multiply and add from the elementary school kids.
Agreed. Humans already implement this: if someone keeps making bigger and bigger unproven claims, eventually they get a reputation as a braggart and a liar, and their claims don’t count for anything.
The point would be that model has to be easily falsifiable—you should be able to see if it doesn’t simulate amoeba.
The Pascal’s mugging is not easily falsifiable. I think of myself as a fairly intelligent individual; it took me 4 days to think that—whoah, the number of possible beings is not as huge as 3^^^^^3 ; if a being is made of 10^30 components each of which can be in either of 10^30 states, the number is only (10^30)^(10^30) which seems way smaller (not sure though, the knuth’s up arrow is unwieldy to compare to anything). And that’s some seriously frigging complex being I’m talking of (to human as human is to amoeba). I can’t rely on myself to falsify such stuff reliably.
You don’t need to, as they would most likely be unable to simulate an amoeba, so you can call their bluff right there.
The issue with pascal’s mugging is that the refusal to simulate amoeba only decreases the probability of validity of the claim by a constant factor, and the mugger can stack more uparrows. Suppose that one in ten thousands superbeings refuses to demonstrate to you the simulated amoeba despite being capable of doing so.
Also, what if it simulated amoeba but claims it’ll torture 3^^^^^3 humans? I can falsify the latter part because there can not be this many distinct human mind states, but only after taking a lot of time to think about the claim.
It seems reasonable to me that my confidence that the mugger can do what it claims to should be inversely proportional to the discrepancy between what the mugger claims and what the mugger is willing to demonstrate. Why do you say it should be a constant factor?
What if one in ten thousands genuine super-beings is not a nice guy and refuses to demonstrate you stuff? The refusal is only linked to ability for those unable to demonstrate, or those barely able to demonstrate. If you mail world’s top mathematicians a question what is 2*3+4 , very few will write you back; you may actually get higher willingness to ‘demonstrate’ ability to multiply and add from the elementary school kids.
True. OTOH, very few of them will approach me with a “mathematician’s mugging,” either.
Agreed. Humans already implement this: if someone keeps making bigger and bigger unproven claims, eventually they get a reputation as a braggart and a liar, and their claims don’t count for anything.