That’s hardly objective. The challenge is to formalize that test.
Btw: the problem you’re having is not due to any decision theory but due to the goal system. You want there to be entertainment and fun and the like. However, the postulated agent had a primary goal that did not include entertainment and fun. This seems alien to us, but for the mindset of such an agent “eschew entertainment and fun” is the correct and sane behavior.
Exactly, though see my comment on a sibling thread.
Out of curiosity though, what is the “Scientology test” ? Is that some commonly-accepted term from the Less Wrong jargon ? Presumably it doesn’t involve poorly calibrated galvanic skin response meters… :-/
That’s hardly objective. The challenge is to formalize that test.
Btw: the problem you’re having is not due to any decision theory but due to the goal system. You want there to be entertainment and fun and the like. However, the postulated agent had a primary goal that did not include entertainment and fun. This seems alien to us, but for the mindset of such an agent “eschew entertainment and fun” is the correct and sane behavior.
Exactly, though see my comment on a sibling thread.
Out of curiosity though, what is the “Scientology test” ? Is that some commonly-accepted term from the Less Wrong jargon ? Presumably it doesn’t involve poorly calibrated galvanic skin response meters… :-/
Not the commenter, but I think it’s just “it makes you do crazy things, like scientologists”. It’s not a standard LW thing.