Branch (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
branchandroot) wrote2012-10-22 10:27 pm
branchandroot) wrote2012-10-22 10:27 pm
Entry tags:
Style and medium are linked, and so is performance
Dear AO3,
I am once again stunned by the sheer ridiculousness of attempting to, what was it now, 'respect everyone's fandom expression' or something when you are not using the system that expression is associated with. To whit: Tumblr style "commentary" tags.
Because here's the thing, dear AO3, Tumblr does not index all those tags. They are not all searchable. And the searchability is determined by the site code, not the site employees. If I type into the tag search field of my dashboard, for example, "lettuce and gravy what", or click on that tag in the post of a moderately horrifying foodstuff, I will get the "no posts found" page, despite the fact that there are clearly posts using that tag. This is because that tag has not been used often enough, and by enough different people, to trigger the completely automatic indexing threshold.
Of course, if I go to the blog of the person who made that entry, and type "lettuce and gravy what" into their blog search field, or click on it while in their blog, I will get all the entries that person may have tagged with that phrase, via a search limited to their own account. This is, naturally, why many people use very idiosyncratic tags on certain types of posts, for things they suspect may be an indexed tag-stream (eg "tenipuri fanning" instead of "Prince of Tennis" on a simple reblog); this prevents a post from being aggregated in the tag-stream, but keeps it searchable on the individual blog.
Furthermore, only the first five tags of any entry are even eligible to be used in aggregating that post into a tag-stream, if one exists. All tags after that may be used to search within that user's blog, but will be ignored when fetching posts to be shown in a "all site posts in this tag" page.
All of this indexing and search scoping is automated in the basic Tumblr code. This is, in a nutshell, why Tumblr has not fallen over and died, suffocated under the weight of unique or rare "commentary" tags. Tumblr fails at documentation, but the basic parameters of how tags are indexed and used to search and present content are solid, and optimized to a high traffic site on which neither permanence nor locating all of a particular kind of content is a major priority.
Tumblr has evolved as a culture of ephemerality, in which commentary is given in the tags and therefore erased with each reblog, in which content is a constantly flowing stream that one watches pass and occasionally dips something out of, in which the only way any user can preserve content for later location is to blog or reblog it on their own account under some internally consistent tag rubric. (And then include the javascript in their theme to actually display their tag list.) I have a hard time imagining anything less suited to the permanent storage and reliable sharing over time of content.
But if AO3 really wants to hold their own funeral and attempt it? Indexing of that field has got to be automated.
I am once again stunned by the sheer ridiculousness of attempting to, what was it now, 'respect everyone's fandom expression' or something when you are not using the system that expression is associated with. To whit: Tumblr style "commentary" tags.
Because here's the thing, dear AO3, Tumblr does not index all those tags. They are not all searchable. And the searchability is determined by the site code, not the site employees. If I type into the tag search field of my dashboard, for example, "lettuce and gravy what", or click on that tag in the post of a moderately horrifying foodstuff, I will get the "no posts found" page, despite the fact that there are clearly posts using that tag. This is because that tag has not been used often enough, and by enough different people, to trigger the completely automatic indexing threshold.
Of course, if I go to the blog of the person who made that entry, and type "lettuce and gravy what" into their blog search field, or click on it while in their blog, I will get all the entries that person may have tagged with that phrase, via a search limited to their own account. This is, naturally, why many people use very idiosyncratic tags on certain types of posts, for things they suspect may be an indexed tag-stream (eg "tenipuri fanning" instead of "Prince of Tennis" on a simple reblog); this prevents a post from being aggregated in the tag-stream, but keeps it searchable on the individual blog.
Furthermore, only the first five tags of any entry are even eligible to be used in aggregating that post into a tag-stream, if one exists. All tags after that may be used to search within that user's blog, but will be ignored when fetching posts to be shown in a "all site posts in this tag" page.
All of this indexing and search scoping is automated in the basic Tumblr code. This is, in a nutshell, why Tumblr has not fallen over and died, suffocated under the weight of unique or rare "commentary" tags. Tumblr fails at documentation, but the basic parameters of how tags are indexed and used to search and present content are solid, and optimized to a high traffic site on which neither permanence nor locating all of a particular kind of content is a major priority.
Tumblr has evolved as a culture of ephemerality, in which commentary is given in the tags and therefore erased with each reblog, in which content is a constantly flowing stream that one watches pass and occasionally dips something out of, in which the only way any user can preserve content for later location is to blog or reblog it on their own account under some internally consistent tag rubric. (And then include the javascript in their theme to actually display their tag list.) I have a hard time imagining anything less suited to the permanent storage and reliable sharing over time of content.
But if AO3 really wants to hold their own funeral and attempt it? Indexing of that field has got to be automated.
no subject
I am pretty fucking tired of people getting on authors' cases for using the site the way they are accustomed to tagging, when the site itself offers no restrictions and no goddamned guidelines anywhere they might come across them.
I do understand people are trying to limit the damage that's being done by the very poor choices that have been made regarding tagging and indexing when they rant about people using the AO3 tags like they're tumblr and attempt to discourage it -- but it doesn't fucking help. They're trying to engineer a cultural solution to an infrastructural problem, which might be possible if the problem were fundamentally created at a reader level, but it's not and it's not going to be solved at the reader level. No, the problem (well, problemS, let's be honest there's more than one when it comes to tagging and indexing) is cultural at a top-level site design and philosophy level and it's going to have to be addressed at that level before we're going to see much benefit.
Which means if you want to get your grump on, get on with the people working on the AO3 and not the people using it.
So, yeah, I'm someone who uses tags on the AO3 occasionally all Tumblr-style. I don't really feel any guilt about this because it's a drop in the fucking bucket, I cannot fix the problems with tagging and indexing just by altering my own tagging practice and I refuse to try. If they're going to allow it, I'm going to tag the way my brain wants to tag.
The thing that's so fucking frustrating is that most of this is at least 70% solvable, if they'd ever accept that they need to declare the goddamned canonicals and publish them already. Even if they did nothing else! Just tell us what the fucking canonicals are! Let us pick from them from whatever, a drop-down, even if/as you still allow us to enter in other forms of tag that you will throw thousands of volunteer hours at wrangling.
(In my wild dreams, I fantasise about them doing that thing we've discussed where they would give us the canonicals and then let us choose our own display text for said, but I'm honestly afraid that site-controlled canonicals + user-controlled display text is a fucking pipe dream.)
Without that, though, I actually think they could do something with the additional tags where they split them into common tags and freeform tags: common tags are canonised and indexed (so, things like: knotting, biting, bdsm, gunplay, etc. would be common tags), and freeform tags are not indexed for general users' search purposes. Instead, they get combed through by the wranglers to see if there's anything that's coming up a lot that needs to be canonised into a common tag. And, yeah, allow the wranglers to associate freeform tags that are semantically equivalent to a common tag so the stuff tagged with the freeform does still pull up when someone does a search on the common tag, sure, but don't actually index the freeform tag, just tag it in the backend with the common tag so it pulls up on the canonical common tag.
But that probably won't ever happen, because it'd mean declaring the goddamned canonicals and publishing them. Because it's okay to have canonicals as long as we never tell the users what they ARE.
no subject
This, this, this. And the Additionals /could/ work on the folksonomy convergence plus unique commentary model if the programming were done to let it happen. Instead we have this half-and-half mess with zero documentation and everyone at the top level trying to CYA like nothing in the whole world is ever their fault instead of, you know, fixing something. Worlds of argh.
no subject
Maybe you're talking about something else, but anybody using the site -- signed-in or not -- can check any tag to see if it's canonical, or if it's associated with a canonical. (Although the word used on the tag pages is "common".)
Examples: Erik Lehnsherr/Charles Xavier is a canonical tag. Erik/Charles is not, and you can see that it's associated with Erik Lehnsherr/Charles Xavier.
no subject
Abracadabra, the canonicals are now visible! They're not highly visible (well, they were always in the drop-downs), but you can find them if you want, and you can see what they attach to. If you use the tag search for a topic and then click on the tag, it now takes you to the landing page for that tag, which shows its relationships with other tags. I've been enjoying myself by cleaning up a bit at the Alternate Universe metatag today. (Fair warning: it needs a lot of help still, so if you have a librarian's soul, you might not want to click that link. Also, just skip the "Tags with the same meaning" box. It's not really important.)
Cheers!