Some questions about tags (let me know if there's a better home for these questions):
1. If people create their own tags, are these visible to everyone, or are they like private tags just for that person? (I assume it's the former; just want to check.)
2. Are there norms around what kinds of tags to create? Should there be?
E.g., should we be careful to avoid a proliferation of extremely fine-grained tags, or just go ahead and not worry?
As case studies, here are some tags I might create if no one tells I maybe shouldn't:
Moral advocacy / values spreading
International relations / war
Maybe most 80k problem areas and career paths not directly covered in the tags already
3. Are there norms around editing tag descriptions? Should there be?
You say "They also now have the ability to edit tag descriptions in a wiki-like fashion", but when someone does something stupid on Wikipedia other people can view the article history and restore old versions. Here it looks like regular users can't do that? That makes me slightly concerned about unilaterally making changes that seem good to me but might seem less good to other people.
As a case study, there are some tags that correspond to topics I've made collections of sources about (e.g., value drift [EA(p) · GW(p)]). I'm inclined to add to the bottom of the description of each of those tags "(Sources on this topic from outside the EA Forum can be found here[link].)"
4. Would it be possible for tag pages to, in future, include links to sources from outside the EA Forum by means other than adding links in the description? Would that be good?
It seems to me that that could be cool, as an elegant way to have the tag page be the ideal go-to link to use, as it'd then have a brief description of the topic and a comprehensive collection of sources.
The goal of tags is to be used as a curated collection over time.
Better to not tag events, as they won't be relevant after a short while.
Better to tag with more specific tags than broad tags.
Vote on tags if you think the post is very relevant to that tag (and so it will show up earlier in the case where people look up through posts on that tag) and downvote when it's less relevant.
Since tags are moderated, it is better to just add many tags and they will be checked for relevance and accuracy. So if you think that some tag would be useful, just go about it. (As a moderator here, I think that tags are very important and I'll gladly spend time going over more tags if that causes more tags to be created)
Good tags should have a balance between not being too small to be irrelevant and not being too big so that the list wouldn't help readers going through it and it would take a lot of overhead to tag new posts.
I don't think that we should be that concerned with having a tag that's too big on the EA forum.
Generally, one can filter through several tags, so I'm not sure what to make of it. @Habryka, I'd be interested in your opinion on this.
Tags should avoid being too near other tags. This can be clarified in the description.
The tagging system is collectively applied which limits its ability to maintain tags with high-degrees of subjective nuance.
Tags overall experience pressure to be as inclusive as possible. If a concept is at all loosely connected to a topic, someone will apply it.
The general result of the above is that a closely related, although theoretically distinct, concepts will end up blurred and having heavy redundant post overlap.
Tag names should be as clear as possible, even for people who don't understand it in full neuance.
It's perfectly fine to use multiple names when is appropriate.
Keep tag names brief. Use & or / instead of and. (I should rename Col≤ctionsandResources).
Tag description should have the tag name in bold and link to related tags. (again, I should change things 😊)
[Low confidence – I'm hashing out my own opinion in public, not trying to apply admin pressure]
I like the tags you've listed there. If you'd asked me to think about concepts in EA and written a (long) list, I'd hope I would have found those. I feel like Political Polarization [? · GW] is maybe more niche than I would do? There's a key difference between us and LW here, which is that LW is investing a large amount of time into creating a whole ontology out of their tagging system, and organizing thing hierarchically, which allows the highlighting of broader tags, while we can't match them in hours devoted if Aaron and I both worked on it full time.
What I've just done is add a tag for most of the shortform collections I'd made that didn't have a tag already. (With a few exceptions where the shortform collection was decently covered by an existing tag, or was really a fairly fuzzy or niche category.)
For some of these, including political polarisation, there aren't many relevant Forum posts I'm aware of. But I felt like maybe that wasn't a big issue, because more posts on the topics might still be created or found later?
And then there's also the issue that some topics might be better off subsumed under something else. That might apply to Political Polarisation (though I'm not sure what it'd be subsumed under?) and to Differential Progress (under Existential Risk or Longtermism (philosophy) or something).
For that reason I didn't (yet?) make Global Catastrophic Risks; that's clearly a non-niche topic, but is maybe covered by the Existential Risk tag (even if the concepts are meaningfully distinct [EA · GW]). And it's why I haven't (yet?) made Meta-Ethics (as it could fit under Moral Philosophy).
Could admins hide or delete tags that they deem overly niche? I'd feel positive about that option being on the table, so people can feel more comfortable about creating tags that might not be worthwhile (which in turn seems good, because many of those tags will indeed be worthwhile). If admins plan to be extremely reluctant to do that, then maybe it'd be good to promote a more cautious norm around tag-creation?
I wish there was a community-led way of deciding about tags. I think LW is making the calls about their tag-classification that they've introduced. (See image.) So maybe it makes sense for us to be more opinionated.
The History tag is for posts that are strongly focused on historical events or trends which don't necessarily connect to other tags (e.g., a post on the history of nuclear weapons [? · GW] should go in that tag instead), or that discuss or make heavy use of historical research methods.
Either way seems ok to me. My thinking was that, if overlap was allowed, a large portion of all posts could be seen as "History" posts. But maybe that's inaccurate or ok. You and other people can feel free to edit any descriptions on tags I made :)
Meta: This discussion makes me realise it's possible it'd be valuable to have some equivalent of Wikipedia's "talk" pages for tags. (But maybe that'd take more work than it's worth.)
ETA: I now think you're right about the History tag, and have adjusted the description accordingly.
You say "They also now have the ability to edit tag descriptions in a wiki-like fashion", but when someone does something stupid on Wikipedia other people can view the article history and restore old versions. Here it looks like regular users can't do that?
My guess is that this is a temporary bug. The History page should allow users to see any previous revisions that were made, and should allow you to compare arbitrary revisions. You can see what it's supposed to look like on LessWrong [? · GW]. With that, restoring previous versions should be pretty easy. I expect that bug will probably be fixed within a week or so, and until then it probably won't be much of a problem.
Is there a way to have footnotes and tables in the same post? I tried just now and can't see a way. (You have to switch to EA forum doc [beta] editor for the tables which kills your footnotes; you have to switch to markdown for footnotes which kills your tables)
edit: I found some markdown code for tables which worked but then had trouble formatting within the table. Decided to just take pictures of the tables instead and upload them as pictures which also works. If anyone knows an easier/nicer way to do this, or if anything is planned, that would be great :)
Thank you for implementing these improvements! I'm particularly pleased to see that I can now add hyperlinks with a shortcut.
A request: not sure if it's related, but around the time these changes were made, whenever I try to open an EA Forum link in a new tab using Vimium (keystroke: capital F), a preview of that page is shown and there's no way I can turn it off without using the mouse. I think the preview should be shown only when one hovers over the relevant link (to inform the user whether the link is in fact worth visiting), whereas in this case it is shown after one selects the link to be opened, so it's serving no useful function. It would be great if this could be fixed.
Huh, no idea why that happens. The hover-previews are not triggered by selection events, but only by the onMouseOver and the onMouseLeave events and have been that way for a long time. My guess is something must have changed in Chrome or maybe in Vimium to make that happen?
Reading through some Github issues for Vimium, it appears that Vimium does indeed send onMouseOver events when clicking on a link, so this is intended behavior as far as I can tell (why I do not know, though I can imagine it overall resulting in a better experience on other websites). I don't currently know how fix this without breaking it on other devices, so I would mostly treat this as a Vimium bug.
Update: In case it is of use to other users of Vimium, I've found a workaround for removing persistent previews. Add this to the 'custom key mappings' field under 'Vimium options':
map x LinkHints.activateMode action=hover
where 'x' is the key binding for this action (I use 'q'). Just press that key and select the link corresponding to the preview you want to disable. (The behavior is a bit erratic; sometimes you need to do it twice.)
A small update by comparison, and it's soon enough after this that I'll post it here:
The 'Forum Favorites' section is now part of a larger Recommendations section. There you'll be able to find your bookmarks, sequences you might want to continue reading, in addition to the Forum Favorites section as before. But unlike before, the posts are randomized, so you won't just see the same 3 posts every time.
It's mostly a UI issue. The comments editor has a lot less space to work with and I haven't yet found a good way to make that UI easily available in the context of comments. You can copy-paste tables and images from the post-editor into the comments editor in the worst-case, which I do recognize is annoying.