I am confused about the quote. Indeed, in that quote Ray is complaining about people tagging things too aggressively, saying basically the opposite of your previous paragraph (i.e. he is complaining that tags are currently often too prominent, look too cluttered, and some users tag too aggressively).
My current sense is that tagging is going well and I don’t super feel like I want to increase the amount of tagging that people do (though I do think much less tagging would be bad).
It’s also the case that tagging is the kind of task that probably has a decent chance of being substantially automated with AI systems, and indeed, if I wanted to tackle the problem of posts not being reliably tagged, I would focus on doing so in an automated way, now that LLMs are just quite good and cheap at this kind of intellectual labor. I don’t think it could fully solve the problem and would still need a bunch of human in the loop, but I think it could easily speed up tagging efficiency by 20x+. I’ve been thinking about building an auto-tagger, and might do so if we see tagging activity drop out of making these buttons less prominent.
(i.e. he is complaining that tags are currently often too prominent, look too cluttered, and some users tag too aggressively).
Right, but the point I was trying to make is that the reason why this happens is because you don’t have sufficient engagement from the broader community in this stuff, so when mistakes like these happen (maybe because the people doing the tagging are a small and unrepresentative sample of the LW userbase), they don’t get corrected quickly because there are too few people to do the correcting. Do you disagree with this?
I think it’s messy. In this case, it seems like the problem would have never appeared in the first place if the tagging button had been less available. I agree many other problems would be better addressed by having more people participate in the tagging system.
I am confused about the quote. Indeed, in that quote Ray is complaining about people tagging things too aggressively, saying basically the opposite of your previous paragraph (i.e. he is complaining that tags are currently often too prominent, look too cluttered, and some users tag too aggressively).
My current sense is that tagging is going well and I don’t super feel like I want to increase the amount of tagging that people do (though I do think much less tagging would be bad).
It’s also the case that tagging is the kind of task that probably has a decent chance of being substantially automated with AI systems, and indeed, if I wanted to tackle the problem of posts not being reliably tagged, I would focus on doing so in an automated way, now that LLMs are just quite good and cheap at this kind of intellectual labor. I don’t think it could fully solve the problem and would still need a bunch of human in the loop, but I think it could easily speed up tagging efficiency by 20x+. I’ve been thinking about building an auto-tagger, and might do so if we see tagging activity drop out of making these buttons less prominent.
Right, but the point I was trying to make is that the reason why this happens is because you don’t have sufficient engagement from the broader community in this stuff, so when mistakes like these happen (maybe because the people doing the tagging are a small and unrepresentative sample of the LW userbase), they don’t get corrected quickly because there are too few people to do the correcting. Do you disagree with this?
I think it’s messy. In this case, it seems like the problem would have never appeared in the first place if the tagging button had been less available. I agree many other problems would be better addressed by having more people participate in the tagging system.