The general tagging principle is that a tag should be added to a post when the post contains a substantive discussion of the tag's topic. As a very rough heuristic, to count as "substantive" a discussion has to be the primary focus of at least one paragraph or five sentences in the post.
I think it'd be good to either:
note that for things like AMAs specifically this should be interpreted as about whether the comments contain a substantive discussion of the tag's topic
note more broadly that substantive discussion of a tag's topic in the comments can qualify a post for that tag
But I'm not sure precisely what we'd want that policy to look like
Maybe we'd want at least 1 sentence in the post as well as at least 5 sentences altogether across the post + comments?
The general tagging principle is that a tag should be added to a post when the post, including its comments thread, contains a substantive discussion of the tag's topic. As a very rough heuristic, to count as "substantive" a discussion has to be the primary focus of at least one paragraph or five sentences in the post or the associated comments.
This assumes we want to use the same heuristic for posts and comments, though your final bullet point seems to implicitly question this assumption.
(If we adopt this revision, other parts of the document may also need to be revised. For example, one can no longer infer an upper bound from the heuristic and the length of the post.)
I do think for most posts (with exceptions for things like AMAs) I'd be less likely to add a tag if there's x amount of discussion of the topic in the comments than if there's x amount of discussion of the topic in the post itself. And I think I'd endorse this. Possible rationales include:
Comments are on average less polished/substantive than posts
People probably read the posts more often than the comments (though I don't know how strong this effect is, and I'm just guessing)
But I think that this doesn't matter a lot, and for simplicity it may be best to not get into that issue in the main guidelines page, since we're just presenting that as a rough heuristic anyway.
I vaguely share your feeling that posts "count for more" than comments, though I can't think of a better heuristic than the one I proposed, so for simplicity I just used the text in my previous comment. Feel free to refine it.
I also removed the paragraph referring to the upper bound, and revised the paragraph that followed it, for unrelated reasons. (I think it's something someone added to the Google Doc I circulated, which I didn't initially read very carefully. As it was worded, the paragraph gave readers advice on how to write posts, rather than on how to tag those posts, which should be the focus of the Tagging Guidelines.)
When a post is tagged with a tag which is a proper subset of another tag, you should add this broader tag only if it is also sufficiently relevant. For example, a post tagged with neglectedness should also be tagged with importance, tractability and neglectedness framework if it also has sufficient discussion of the ITN framework as such, rather than just of neglectedness specifically.
I'm not sure I understand this, or maybe I'm not sure I understand what the problem these sentences are aiming to solve is. What about, for example, a post that's entirely about 1 aspect of AI alignment/governance for which we have a specific tag? I think it should also get the AI alignment/governance tag, even if every paragraph is about that 1 aspect, because it would be good for the post to be findable via those broader tags. It would be a bit odd for more focused posts to be excluded from those broader pages even in cases where they're great, have high karma, etc.?
(But mostly these guidelines look great, as we discussed earlier, and thanks for making them!)
The gist of what I meant is just that the parent tag has to meet the same relevance standards as any other tag, and not be included merely because of its logical relation to its child. I didn't mean to suggest—though the passage you quote is worded in a way that seems to make that suggestion—that unless the post covers material specific to the child tag, its parent should not be tagged. I have revised the text to make this clearer (and fixed the link you mentioned in the other comment—thanks).
Check that the tag you would like to create does not exist under a different name. You can see a list of all tags here.
That link goes to the tagging dashboard. I imagine it's meant to go to the Wiki homepage (formerly the tag portal)? But I'm unsure, since I think the Wiki homepage often doesn't list all tags - there've been a few times I've tried to make a tag but then discovered either that one or a related one already existed, could've been found via the search bar, but weren't on the Wiki homepage. (But maybe that's changed now.)
So I think the link should be changed to the Wiki homepage + the phrasing should be something like "You can use the search bar to see if a tag with a given name exists, or see a list of all tags here (except possibly new tags that haven't yet been added there)." But I'm not sure.
The auto-generated alphabetical list in the Wiki homepage (after the list organized thematically) should be exhaustive. Did you notice any missing entries from that list?
I'm almost certain there've been times when I used command+f to search something on that homepage, nothing came up, but then when I made the tag or searched for it it turned out it existed. But maybe this just used to be the case and isn't anymore, or something.
I think it'd be good to either:
Maybe
This assumes we want to use the same heuristic for posts and comments, though your final bullet point seems to implicitly question this assumption.
(If we adopt this revision, other parts of the document may also need to be revised. For example, one can no longer infer an upper bound from the heuristic and the length of the post.)
I think that that principle sounds good to me.
I do think for most posts (with exceptions for things like AMAs) I'd be less likely to add a tag if there's x amount of discussion of the topic in the comments than if there's x amount of discussion of the topic in the post itself. And I think I'd endorse this. Possible rationales include:
But I think that this doesn't matter a lot, and for simplicity it may be best to not get into that issue in the main guidelines page, since we're just presenting that as a rough heuristic anyway.
I vaguely share your feeling that posts "count for more" than comments, though I can't think of a better heuristic than the one I proposed, so for simplicity I just used the text in my previous comment. Feel free to refine it.
I also removed the paragraph referring to the upper bound, and revised the paragraph that followed it, for unrelated reasons. (I think it's something someone added to the Google Doc I circulated, which I didn't initially read very carefully. As it was worded, the paragraph gave readers advice on how to write posts, rather than on how to tag those posts, which should be the focus of the Tagging Guidelines.)
I'm not sure I understand this, or maybe I'm not sure I understand what the problem these sentences are aiming to solve is. What about, for example, a post that's entirely about 1 aspect of AI alignment/governance for which we have a specific tag? I think it should also get the AI alignment/governance tag, even if every paragraph is about that 1 aspect, because it would be good for the post to be findable via those broader tags. It would be a bit odd for more focused posts to be excluded from those broader pages even in cases where they're great, have high karma, etc.?
(But mostly these guidelines look great, as we discussed earlier, and thanks for making them!)
The gist of what I meant is just that the parent tag has to meet the same relevance standards as any other tag, and not be included merely because of its logical relation to its child. I didn't mean to suggest—though the passage you quote is worded in a way that seems to make that suggestion—that unless the post covers material specific to the child tag, its parent should not be tagged. I have revised the text to make this clearer (and fixed the link you mentioned in the other comment—thanks).
That link goes to the tagging dashboard. I imagine it's meant to go to the Wiki homepage (formerly the tag portal)? But I'm unsure, since I think the Wiki homepage often doesn't list all tags - there've been a few times I've tried to make a tag but then discovered either that one or a related one already existed, could've been found via the search bar, but weren't on the Wiki homepage. (But maybe that's changed now.)
So I think the link should be changed to the Wiki homepage + the phrasing should be something like "You can use the search bar to see if a tag with a given name exists, or see a list of all tags here (except possibly new tags that haven't yet been added there)." But I'm not sure.
The auto-generated alphabetical list in the Wiki homepage (after the list organized thematically) should be exhaustive. Did you notice any missing entries from that list?
I'm almost certain there've been times when I used command+f to search something on that homepage, nothing came up, but then when I made the tag or searched for it it turned out it existed. But maybe this just used to be the case and isn't anymore, or something.
Good to know. If you notice this again, could you please let me know so I can investigate? Thanks.
Sure thing.