I’m skeptical of any process where, as appears to be the case here, calculating expected values is demoted to a weak tiebreaker. His description implies (though not explicitly) that expected values are only considered heavily in their calculations when comparing within a domain.
I’ve only breezed through Holden Karnofsky’s full article, but the demotion of calculated expected values only seems to apply where Knightian uncertainty is considerable.
I’m skeptical of any process where, as appears to be the case here, calculating expected values is demoted to a weak tiebreaker. His description implies (though not explicitly) that expected values are only considered heavily in their calculations when comparing within a domain.
I’ve only breezed through Holden Karnofsky’s full article, but the demotion of calculated expected values only seems to apply where Knightian uncertainty is considerable.
That’s his assertion but his examples don’t really seem to support it. Hence my skepticism.