Part of the idea of curation is that some posts are what one might call Evergreen. They make sense out of the context of the discussion at that time, or are part of a full Evergreen discussion that makes sense out of the context of that time. Also, some posts are designed largely as exercises or places to sort things out, versus creating Evergreen things that last.
This especially applies to calls to action that no longer make any sense given the time that has passed.
If we’re going to do recommendations as the top thing on the page every time, it seems like it would be worth it to remove the ones that are about topics that no longer apply or make sense. I realize this will involve judgment calls. I don’t have a good solution beyond ‘someone goes through them and picks which ones not to include.’
Current algorithm is: The LessWrong team is removing all the ones that don’t seem evergreen as soon as we see them in the recommendation section.
Likely also makes sense to do some systematic passthrough. My guess is there are around 3000 posts that can be recommended. Going through all of them is some work, but not prohibitively much work.
Part of the idea of curation is that some posts are what one might call Evergreen. They make sense out of the context of the discussion at that time, or are part of a full Evergreen discussion that makes sense out of the context of that time. Also, some posts are designed largely as exercises or places to sort things out, versus creating Evergreen things that last.
This especially applies to calls to action that no longer make any sense given the time that has passed.
If we’re going to do recommendations as the top thing on the page every time, it seems like it would be worth it to remove the ones that are about topics that no longer apply or make sense. I realize this will involve judgment calls. I don’t have a good solution beyond ‘someone goes through them and picks which ones not to include.’
Current algorithm is: The LessWrong team is removing all the ones that don’t seem evergreen as soon as we see them in the recommendation section.
Likely also makes sense to do some systematic passthrough. My guess is there are around 3000 posts that can be recommended. Going through all of them is some work, but not prohibitively much work.