Humans, and ‘the factors we deem relevant to that decision’, are complicated enough that such things change pretty regularly—sometimes on a moment to moment basis.
This post is, among other things, an attempt to make one of the causes of such changes apparent, so that a particular class of changes can be accurately predicted.
It’s probably not useful to you, at least in the way that Eliezer intended it to be useful, because your mind probably doesn’t work that way. It’s still useful to some significant portion of the rest of us.
To the extent that humans are complicated enough to change unexpectedly on a moment to moment basis in that manner, the changes are not regular.
To the extent that a human’s decision theory or value change irregularly, any talk of optimization is moot, as irregularity necessarily leads to exploitable inconsistency at some level.
Humans, and ‘the factors we deem relevant to that decision’, are complicated enough that such things change pretty regularly—sometimes on a moment to moment basis.
This post is, among other things, an attempt to make one of the causes of such changes apparent, so that a particular class of changes can be accurately predicted.
It’s probably not useful to you, at least in the way that Eliezer intended it to be useful, because your mind probably doesn’t work that way. It’s still useful to some significant portion of the rest of us.
A fair point, but:
To the extent that humans are complicated enough to change unexpectedly on a moment to moment basis in that manner, the changes are not regular.
To the extent that a human’s decision theory or value change irregularly, any talk of optimization is moot, as irregularity necessarily leads to exploitable inconsistency at some level.
Humans change their minds less often than they think.
I rarely change my mind too, but estimate this frequency more accurately.