Should rationality be determined by maximizing expected value or by maximizing the median or by maximizing the mode?

exercise: how much would someone have to pay you to cut off your arms? would you make a bet where: if you win you get 100x that amount, if you lose your arms are cut off, and the chance of winning it is 1 in 50?

i contend that you are influenced by not just the expected value, but also the fact that there is a 98% chance that you will end up with no arms and nothing to show for it.

i contend that people are mainly influenced by expected value when there is the possibility of iteration so that on timescale on the order of their lifetime, there is a decent chance of winning --- otherwise, the expected value will have to be ridiculously large in order to be decisive.


"Aumann's agreement theorem says that two people acting rationally (in a certain precise sense) and with common knowledge of each other's beliefs cannot agree to disagree. More specifically, if two people are genuine Bayesian rationalists with common priors, and if they each have common knowledge of their individual posterior probabilities, then their posteriors must be equal." -- [1]

" kybernetikos 2 days ago [-]

If they have common priors which is almost never the case.

Most disagreements are to be found in the priors, which is why defining terms and trying to agree what the relevant priors are can be very useful.

reply " -- [2]

" The Ideological Turing Test is a concept invented by Caplan to test whether a political or ideological partisan correctly understands the arguments of his or her intellectual adversaries. The partisan is invited to answer questions or write an essay posing as his opposite number; if neutral judges cannot tell the difference between the partisan's answers and the answers of the opposite number, the candidate is judged to correctly understand the opposing side. " -- [3]


todo should combine the pages [4], [5], [6], [7], [8].