Independent alignment researcher trying to build some philosophy in public
berns
Karma: 17
Verify, but Trust
I’ve updated the footnote to specify that I mean commitments about our future behavior which typically, as a consequence, narrow our future options. I think that’s basically just the same as saying a commitment about future behavior that carries any weight at all. Yes, I would say that the reason FDT solves Parfit’s hitchhiker is exactly because of its greater power to precommit than e.g. CDT (which is basically unable to precommit at all.) Thinking about a value like honesty with a quantifiable dollar figure doesn’t really reflect our behavior or desires about it. Treating honesty like a numerical preference basically forces it to be nothing more than a scalar input to a calculation you’re always running. A big part of the point of the Mary/Jane example is that that’s a bad idea when it comes to accounting for values you actually care about.
“Dominant doesn’t mean perfect”
Hello,
I’m posting this here, but I’m not sure if there’s a better way to escalate stuff like this in the future. My recently posted first post on Precommitments was categorized by the autoclassifier as a Personal Blogpost, along with a message that the classification would be reviewed by a human within ~24 hours. As far as I can tell, the post meets all of the descriptions for a Frontpage post (“on topic” for the core interests of LW, explanatory rather than argumentative) and none of the ones for a Personal Blogpost (niche, meta, “personal ramblings.”) The post has been up for more than 48 hours and I didn’t get a response through the mod chat widget to two messages. Thanks for any help on this and sorry if this was not the right place to put it.
Good suggestion, thanks