Work Text:
TOASTYSTATS: AO3 TAG SEARCH.
Hey all! It’s been a while since I’ve had time to do any fandom stats. I’m stretching my muscles and dusting off my spreadsheets by taking a look at a few quick things.
Based on my searches above, it seems like AO3 users have created about 1.9 million tags. Dude. That’s pretty neat. :) So, first up: what we can learn about these tags from the AO3 Tag Search tool. (I’m doing this in part because I think most AO3 users might not know about this cool feature!)
Some background on tags
When AO3 authors & creators upload a new fanwork, they get to add tags via the following form:
As you can see, there are four different types of tags where creators can type in anything they want -- for the tags that describe fandom, relationship, character, and any additional tags (also known as freeforms -- these are tags like “Angst” and “Alternate Universe”).
Because creators can describe the same thing in lots of different ways -- e.g., they might type Laura/Carmilla, Carmilla/Laura, Laura Hollis/Carmilla Karnstein, Hollstein but all be referring to the same relationship -- AO3 tag wranglers wrangle tags together and declare a particular tag the canonical version (and the other versions synonyms):
(There’s also a hierarchy of sub and meta tags.)
So I was curious -- how many of the tags created on AO3 are freeform (Additional Tags) compared to other categories? And how many of those get made canonical?
On to the data...
To get the first graph above (reprinted here for easy reference), I did a tag search where I selected each subcategory (e.g., “fandom”) and hit search, and recorded the number of results that were returned:
Keep in mind -- this includes both canonical and non-canonical tags; basically, every unique string an AO3 user has typed into one of these fields.
This general order makes sense to me -- there are far fewer fandoms than there are characters in the fandoms, e.g. And it makes sense that there are more combinations of characters (relationships) than individual characters. (Though not very many more, which is interesting to ponder.) And of course, I’d expect more freeform/additional tags than the other types.
But it’s interesting to see how many -- or, perhaps, how few -- freeform tags there are. On the one hand, there are over a million freeform tags. On the other hand, while AO3 users have indeed generated a whole lot of freeform tags, that’s only about three times as many as the number of relationship or character tags. Sometimes people complain about how horribly chatty AO3 tags can be, but I think it’s actually pretty reasonable that in describing tropes, genres, AUs, kinks, canon timing/context, etc., there would be substantially more freeform tags. This is fewer than I expected, actually.
It’s fun also to look at some examples of how users are creating and using tags by browsing through some of the results returned by doing a tag search for freeform tags -- the bold ones are canonical, and the non-bold ones are not:
Even the non-canonical tags aren’t necessarily chatty or gibberish; they are often tags like crack!fic that many users have used -- there’s just an alternate spelling that AO3 has chosen as the canonical one.
I recommend browsing a bunch of pages of freeform tags if you’re interested in quickly seeing lots of examples people are using tags and how AO3 is canonicalizing them (though I don’t understand how the results are sorted, but it’s not random, and the number of canonical tags decreases as you go on).
Okay, so if there are nearly 2 million tags and a small corps of AO3 tag wranglers, how many tags actually get made canonical? Let’s return to the second graph:
AO3 makes hundreds of thousands of tags canonical -- amazing! That’s a great deal of review and curation the volunteers are doing so we can find the fanworks we seek. Thanks, wranglers!
Let’s look at this again as percentages of each type:
33% of fandom tags, 38% of character tags, and 37% of relationship tags are made canonical. Meanwhile, only 3% of freeform tags are canonical. There could be a lot of reasons for the large difference, including:
- There are way more freeform synonyms than synonyms in the other categories, so far fewer of these should be canonical. E.g., look at all the synonyms for Alpha/Beta/Omega Dynamics:
- AO3 tag wranglers prioritizing fandom, character, and relationship tags because most fans rely on those most to find stories.
- It’s a lot easier to wrangle those categories -- you can usually quickly tell which character/fandom/ship a creator is referring to, because there is a fairly fixed set of options. Freeforms can be literally anything.
- There are just SO MANY more freeforms.
- I believe AO3 is planning to redo aspects of their freeform wrangling hierarchy and may no longer be creating very many new canonical freeform tags in the interim. (Tag wranglers/AO3 staff, please correct me if that’s wrong.)
It’s probably a combo of several of these reasons.
...Okay, this has been my attempt at a “super quick” look at some fandom stats, but it took me (and my cats) a few hours anyway. So I’m going to quit here and release this into the wild now. But I’d love to hear other thoughts and feedback, as always. And here’s the raw data, if anyone wants it!