roystgnr comments on Don't Get Offended - LessWrong

32 Post author: katydee 07 March 2013 02:11AM

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

Comments (588)

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

Comment author: roystgnr 14 March 2013 07:08:46PM *  1 point [-]

The problem with caching is just that sometimes the cache falls out of sync; you want to evaluate some complex problem f(x), and if you've previously evaluated some similar f(y) it's faster to evaluate "if (resembles_y(x)) then cached_f(y) else f(x)", but if resembles_y(x) isn't precise enough, then you've overgeneralized.

But the correction "Stifle it" doesn't seem to be pinpoint-precise either, does it? It's an overgeneralization that just generalizes to the opposite conclusion.

If you don't want people to overgeneralize, then you have to be specific - "in cases where A, B, or C hold, then you want to avoid giving offense; if D, E, or F hold then giving offense may have higher utility", etc - and just trying to begin nailing down this kind of precision is likely to require hundreds of comments, not just a couple sentences.