You're looking at Less Wrong's discussion board. This includes all posts, including those that haven't been promoted to the front page yet. For more information, see About Less Wrong.

PhilGoetz comments on Value evolution - Less Wrong Discussion

14 Post author: PhilGoetz 08 December 2011 11:47PM

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

Comments (111)

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

Comment author: PhilGoetz 09 December 2011 03:29:21AM *  3 points [-]

Do you grant that I can have reflective preferences about the way my values should change in the future?

No. If you have a preference about how your values should change, it means you have conflicting values. If you think that you want your values to change, this probably means that the conscious you places more value on one preference, and the unconscious you places more value on another. This is what is happening when people say they wish they could eat less. Their minds want to eat less, and their bodies want to eat more.

Comment author: roystgnr 09 December 2011 07:43:45PM 0 points [-]

You and paulfchristiano seem to be using the word "way" in two different ways.

Your post makes sense if I replace "the way" and "how" with "the direction in which". His makes sense if I replace "the way" by "the means with which".

To apply your example: I can't consistently prefer a value change like "I should eat more fish", because if I wholeheartedly preferred that then I'd already be eating more fish. I can prefer a value change like "I should eat more of whatever foods are recommended by good nutritional studies that I haven't seen yet", because although I cannot identify any specific failing of my current values I can identify that there are specific ways in which they might be improved in the future by unexpected new information.

This possibility of improvement applies only to instrumental values and self-inconsistent terminal values, but that's still pretty useful. How many people currently have and can unambiguously define self-consistent terminal values?