Posts

References that treat human values as units of selection? 2019-06-09T18:04:09.074Z · score: 13 (4 votes)

Comments

Comment by milan-cvitkovic on References that treat human values as units of selection? · 2019-06-10T20:11:21.704Z · score: 1 (1 votes) · LW · GW

Thanks, Charlie!

Modifying yourself to want bad things is wrong in the same sense that the bad things are wrong in the first place...

I definitely agree with this, and have even written about it previously. Maybe my problem is that I feel like "find the best values to pursue" is itself a human value, and then the right-or-wrong-value question becomes the what-values-win question.

Comment by milan-cvitkovic on References that treat human values as units of selection? · 2019-06-10T00:22:16.063Z · score: 2 (2 votes) · LW · GW

Thank you! Carl Shulman's post still seems written from the some-values-are-just-better-than-others perspective that's troubling me, but your 2009 comment is very relevant. (Despite future-you having issues with it.)

The question "which values, if any, we should try to preserve" you wrote in that comment is, I think, the crux of my issue. I'm having trouble thinking about it, and questions like it, given my "you can't talk about what's right, only what wins" assumption. I can (try to) think about whether a paperclip maximizer or The Culture is more likely to overrun the galaxy, but I don't know how say that one scenario is "better", except from the perspective of my own values.

Comment by milan-cvitkovic on Is value drift net-positive, net-negative, or neither? · 2019-05-27T02:03:46.096Z · score: 1 (1 votes) · LW · GW

I would argue that the concept of value drift (meaning "a change in human values from whatever they are currently") isn't really sensible to talk about. Here's a reductio argument to that effect: Avoiding bad value drift is as important as solving value alignment: https://docs.google.com/document/d/1TDA9vHBT7kN9oJ69-MEtbXZ_GiAGk3aRhLXFgkLv8XM/edit?usp=sharing

It's hard to compare values on their "goodness". I prefer to think of them as phenotypes and compare them on their adaptive benefits to agents that hold them. After all, it doesn't really matter what's right: it matters what wins.