Eugine_Nier comments on Values vs. parameters - Less Wrong

9 Post author: PhilGoetz 17 May 2011 05:53AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (65)

You are viewing a single comment's thread. Show more comments above.

Comment author: Vaniver 17 May 2011 12:54:55PM 2 points [-]

It seems to me that eliminating space-discounting is done for memetic, not genetic, reasons. And, as the triumph of individualistic nations over collectivist tribes shows, it outcompetes. "I trust strangers in the way a hunter-gatherer trusts his cousins" is not just a platitude for making people think you're moral, but also a good strategy for conquering the world.

Do values come from genes, and parameters from memes? The other way around? A mix? My guess is that most of our value drift will represent genetic strategies being replaced by memetic strategies, and so talking about values and parameters may not make a whole lot of sense.

It seems to me that time discounting is a heuristic to deal with death- if that issue gets resolved, then you don't need to discount as heavily. I'm not sure how a utility function would deal with memories, though- if I get to have an awesome experience at one point in my life, it seems like the earlier the better, as I can only remember the experience after I have it (I can anticipate it beforehand, but I imagine that'll be weaker). So my guess is immortals (or, at least, billion-year lifespan individuals) will have a weak preference for the present.

Comment author: Eugine_Nier 17 May 2011 03:01:36PM 3 points [-]

It seems to me that time discounting is a heuristic to deal with death

I think a large part of time discounting is dealing with uncertainty about the future. It's notable that people in unstable societies/situations tend to time discount more then people in more stable situations.

Comment author: PhilGoetz 17 May 2011 06:26:48PM 0 points [-]

See Eliezer's post against discounting - I've been trying to address the problem that's left even after you factor out uncertainty.

Comment author: timtyler 17 May 2011 06:46:29PM 1 point [-]

Note that instrumental discounting is not just down to uncertainly - there's also mortality, future-impotence, resource limitations, the divide-and-conquer strategy - and so on. You should probably be trying to factor all of those things out.