I think there are (almost?) always multiple reference classes, outside views, or base rates one could use, and that one (almost?) always has to get at least a bit inside-view-y when deciding how much weight to give each (or just which to include vs exclude). This also leads me to strongly prefer the term “an outside view”; I think “the outside view” implies we’re facing a simpler situation than we’re ever actually facing.
(I definitely do think considering outside views is very useful, though!)
Some relevant posts on this issue, in descending order of relevance:
Interesting comment.
I think there are (almost?) always multiple reference classes, outside views, or base rates one could use, and that one (almost?) always has to get at least a bit inside-view-y when deciding how much weight to give each (or just which to include vs exclude). This also leads me to strongly prefer the term “an outside view”; I think “the outside view” implies we’re facing a simpler situation than we’re ever actually facing.
(I definitely do think considering outside views is very useful, though!)
Some relevant posts on this issue, in descending order of relevance:
https://www.lesswrong.com/posts/SxpNpaiTnZcyZwBGL/multitudinous-outside-views
https://www.lesswrong.com/posts/iyRpsScBa6y4rduEt/model-combination-and-adjustment
https://forum.effectivealtruism.org/posts/SBbwzovWbghLJixPn/what-are-some-low-information-priors-that-you-find