Is anyone interested in giving this a second nomination?
I argue that this post is significant for filling in a gap in our canon: in “Where to Draw the Boundary?” (note, “boundary”, singular), Yudkowsky contemptuously dismisses the idea that dolphins could be considered fish. However, Scott Alexander has argued that it may very well make sense to consider dolphins fish. So … which is it? Is Yudkowsky right that categories must “carve reality at the joints”, or is Alexander right that “[a]n alternative categorization system is not an error, and borders are not objectively true or false”?
In this post, “Where to Draw the Boundaries?” (note, boundaries, plural), I argue that Yudkowsky is right that categories must carve reality at the joints; however, I reconcile this with Alexander’s case that dolphins could be fish with a simple linear-algebraic intuition: entities might cluster in a smaller subspace of configuration space, while failing to cluster in a larger subspace. Clusters in particularly “thin” subspaces (like a fake job title that nevertheless makes predictions on the “what’s printed on business cards” dimension) may fail to be useful.
If people are wary of the political context in which this was written, I put up a non-Frontpage containment thread in case anyone wants to complain or ask questions there.
I’m expecting to probably nominate this, but first want to re-read both the post and it’s predecessors and think about it a bit.
(My recollection is that I wasn’t bothered by the political context, but feel like the post is a bit confusingly structured and I would probably recommend a significant rewrite to make it’s point more clear and more clearly motivated. It takes a long time before the post gets to a point where I understand why I might care about any of this. I think the fake-job-titles is actually a pretty good example without being especially controversial, which should maybe be more front-and-center?)
Is anyone interested in giving this a second nomination?
I argue that this post is significant for filling in a gap in our canon: in “Where to Draw the Boundary?” (note, “boundary”, singular), Yudkowsky contemptuously dismisses the idea that dolphins could be considered fish. However, Scott Alexander has argued that it may very well make sense to consider dolphins fish. So … which is it? Is Yudkowsky right that categories must “carve reality at the joints”, or is Alexander right that “[a]n alternative categorization system is not an error, and borders are not objectively true or false”?
In this post, “Where to Draw the Boundaries?” (note, boundaries, plural), I argue that Yudkowsky is right that categories must carve reality at the joints; however, I reconcile this with Alexander’s case that dolphins could be fish with a simple linear-algebraic intuition: entities might cluster in a smaller subspace of configuration space, while failing to cluster in a larger subspace. Clusters in particularly “thin” subspaces (like a fake job title that nevertheless makes predictions on the “what’s printed on business cards” dimension) may fail to be useful.
(Perhaps also of significance is that the job title example from this post ended up inspiring the local work on simularcum levels.)
If people are wary of the political context in which this was written, I put up a non-Frontpage containment thread in case anyone wants to complain or ask questions there.
I’m expecting to probably nominate this, but first want to re-read both the post and it’s predecessors and think about it a bit.
(My recollection is that I wasn’t bothered by the political context, but feel like the post is a bit confusingly structured and I would probably recommend a significant rewrite to make it’s point more clear and more clearly motivated. It takes a long time before the post gets to a point where I understand why I might care about any of this. I think the fake-job-titles is actually a pretty good example without being especially controversial, which should maybe be more front-and-center?)
I’m hoping the sequel (forthcoming later this month) will be a lot clearer!