Femslash week in review: 2019-03-12 - 2019-03-18

High-level stats for week of 2019-03-12 - 2019-03-18

• Total works categorized F/F on AO3: 3898 (-165 from last week)


• Works I classified F/F: 2349 (-80 from last week) (1042 new, 1307 continued)


• 0.94% of all 250023 AO3 works I've classified F/F were updated this week

A few callouts this week:

• Okay, folks, we're back. What's new? I feel like Critical Role has been in the top 20 more recently than had previously been the case, but I also feel like the ratings in the lower 10 are pretty flat, so anything could jump into the top 20 somewhat arbitrarily.


• But let's be serious for a moment: it's going to be the Carol/Maria show here at the Week In Review for the next while, so everyone remain calm. Have some recs from this week: two primarily set immediately post-canon here and here; one primarily backstory; one cute post-canon (warning: lots of conversations with dudes).


• Also a cute Carol/Minn-Erva short piece.


• Maybe y'all want to know a little more about why I took a week off? As it turns out, that's related too: last week, I was running behind on everything fandom-related because life, so it was the last minute when I needed to be posting, and I figured I'd just pull up the list of Carol/Maria fics from my recs engine, and give a run-down ...and then there were no Carol/Maria listings in my recs engine. Which seemed improbable. I figured out why the Carol/Maria tag had been misclassified reasonably fast, but then I failed to fix it immediately due to a mistake, and broke some of my graphs in the process because of another minor bug. So I would have had to sort that out to get anything postable at all, and I didn't want to post without fixing the initial problem because I wanted to know if MCU was going to get into the top 20 (which I'm still not sure about, because last week's stats are still a little messed up, but I'm not going to worry about it at this point). And I was just out of time, so I decided to punt, which, thanks for understanding, all.
  • What was the bug, though? When I query ship tags from AO3, one of the things I look for is whether they are listed as common tags or not. I then assume that "uncommon" tags (the ones that show up in the AO3 tag search UI as something like "This is not a common tag, and cannot be filtered on") can't be F/F ships. The thinking is that if all the AO3 tag-wranglers don't have time to fix up a particular tag, I probably don't have time to do it myself working single-handledly. That's all fine and well; the problem is that I only check whether a tag is uncommon the first time I look it up. So if I happen to look up a tag right after the first time someone uses it when it's still unaudited by the tag wranglers, it doesn't matter how big it gets later, my stats will exclude it forever. Whoops.
  • How am I fixing this root cause? Well, I haven't yet, but auditing ship tags has become my new top priority. I started out by doing some low-hanging fruit auditing; trying to find duplicate ship tags in my cache, and I've been using that as a first pass of just cleaning things up a bit. I think the right fix for the uncommon tag problem, is any week a ship appears in new works, look it up again even if I already have it marked uncommon. Ships which are actively appearing in works are the ones which are most likely to have been revisited by AO3. But I'll want to do a one-time audit as well: at this moment, I have 134585 ships in my cache, and 34447 of them are marked uncommon, which presumably means I'm missing enough data to make a difference.
  • At the risk of sounding like I'm making excuses for myself: working with AO3's tags is a messy business. The programmers, tag wranglers, and participants do an excellent job of keeping the data rich and usable, but there's just a lot of distance between the data as it's available, and being able to provide meaningful numbers about which fandoms are big, which ships are big, and which fandoms are "ensemble-y", while cranking out new numbers in a reasonable amount of time/effort every week, and sticking with my stated goals of excluding cis dude and RPF ships. On the one hand, this week's bug exposed another gap which should make people nervous if they are counting on these stats for anything important. (Umm, which, maybe don't?) On the other hand, I actually don't feel like my data is that much of a mess, all things considered, and hopefully I'll be able to learn from issues like this and continue to make progress making it more accurate and useful.
  • Finally: thanks everyone who left me nice comments and likes on my "taking a week off" post! I'm always pleased to see folks are reading the stats. Another thing I'm going to prioritize in the near future is doing a reader survey to find out what y'all want to see more (or less) of; the amount of work needed to patch known data accuracy bugs means I want to take a break and find out what people actually care most about, as readers.

( Full top-20 table and description of methodology after the jump )