pondering fanfiction numbers

[alias_sqbr]

I argued with someone on tumblr who said the quantity of fanfic produced has starkly decreased since 2009, but while I was sure they were wrong I couldn't back myself up with hard evidence. Luckily their main argument was "these days there's nothing as active as Stargate Atlantis fandom used to be", which was pretty easy to refute.

Beyond that specific conversation, I'm wondering if there's been any studies of fanfic/fanwork volume over time, or if there's any way to estimate even a rough ballpark. It would be interesting to even just see changes within a specific community over time, eg Jane Austen fandom after various adaptations. Or comparing numbers within different archives, like fanfiction.net vs AO3. But the AO3 is the only archive I've come across which makes it easy to look back on past numbers for that sort of thing, and it's too young and unevenly popular to be a reliable sample.

I find poking at these sorts of numbers fun, I think I might do it some more. Fans constantly make sweeping statements with zero intellectual rigour. And intellectual rigour is important, dammit! Plus it's fun to just ask an interesting question and find out the answer. I appreciate it when people with a history/media studies etc degree add some Actual Facts to these conversations, and I feel like there's numerical data holes I could fill. For a start, noone seems to be very interested in analysing femslash and fanart, which are two things I'm into and could poke at.

I have massive imposter syndrome about whether I'd actually add anything to the conversation, though, and about my qualifications to do this sort of analysis accurately. I studied just enough stats at uni to be aware of how much stats I don't know, and my training is all in abstract mathematics etc, not...sociology?? I don't even know the right discipline. But I do have a computational mathematics phd and worked as a data analyst for a while, which is a hell of a lot more qualification than I have to analyse media etc, which I certainly pontificate about plenty. I guess since I haven't studied any humanities at all it's easier to coast by on a comforting wave of Dunning Kruger ignorance of my ignorance.

Anyway, even if I just have fun and noone else finds this stuff interesting or useful, it's just nice to feel interested in doing something vaguely mathsy again. The awfulness of my phd followed by declining health getting me fired from a sequence of science related jobs, combined with brainfog, has rather sucked the joy out of maths and science for me for the last ten years or so, when it used to be what I intended to devote my life to.

If I do end up poking at fandom stats some more I should look into existing methodoloogy for analysing social group numbers methodically. Not that they gave me any training in that sort of thing when I crunched medical data for the Health department, but back then I wasn't the one deciding what numbers to crunch. As it happens a friend with a sociology degree is coming over tonight, I could ask him, though he can be a bit Smug Mansplainy Arts Graduate.

So, things to do:
Read through the links on this post
Refresh my decades old statistics knowledge
Have a proper look at what toastystats is and isn't doing, since she seems to be the most visible person analysing this stuff at the moment.

Actually one thing I might be good at is writing up methods so other people can do their own analysis. If I learned one thing during my maths phd it's that I am, at best, a middling mathematician, but I'm not bad as a science communicator.

Also: while I was looking around for the tumblr thing I came across the fact that about every second fanfiction.net story ends up deleted??

Also, I love the fandom choice in the sentence "Now, it might not seem substantial, but Twilight with its 150,000 uploads may have up to 2000 dead stories counted as alive every day"

Comments

[merit]

If you watch the slow spiralling down of a fandom, things will look less active. Using a show that ended ten years ago is perhaps not the best comparison. I sometimes look at the top fandoms for various categories on Ao3 and some I've barely heard about but the numbers of fics uploaded don't lie.

I'd be curious about longfic stats over the last decade or so but once again easy comparisons would be difficult.

[alias_sqbr]

Right! And yeah, same.

[ursula]

I've seen interesting fandom stats lately from franzeska and letzan. There are interesting questions here about whether to sample or suck in as much as possible, how to clean data, etc.

[alias_sqbr]

Tanks, I'll check them out!

[lea_hazel]

As a non-mathy person I just don't understand how it's possible to even approximate how much fanwork is being produced at any given time, which is why I have trouble taking such claims seriously.

I've encountered them before, but they were phrased so poorly... I'm very wary of any claims of "fans these days are fanning wrong".

[alias_sqbr]

I feel like maths is actually the wrong discipline for that sort of question, while you would use maths in the calculation I feel like historians are more the sort of people used to figuring out this sort of thing. So I have no idea how feasible it would be to figure out.

But I'm pretty sure neither do any of the folk making sweeping pronouncements about it >.>

[lea_hazel]

Right. "Internet historian" is probably a discipline all its own.

[letzan]

Hi! I compile fandom stats and I can't even remotely answer any of your questions, so how's that for useful?

The problem is that the kind of data-scraping-based analysis I do starts with the assumption that you already know what is and isn't in your dataset. So I entirely agree with your other commenters who are saying that you need a historian not a statistician. (I'm neither, but much more like the latter if I have to pick.)

I can tell you lots of stuff about AO3 femslash over time (which is what my stats are about), but the primary thing I can say about AO3 in 2009 is that it barely existed yet. So I've made the deliberate decision that I don't really need to be able to compare anything happening right now to Xena in its heyday, much less Sailor Moon. My perception from the data I have is that femslash on AO3 is exploding. I suspect this is largely because AO3 usage overall is exploding, and only secondarily because people have gotten more interested in female fictional characters. But those two changes would look the same to me, making my stats even less useful for answering your question.

Just to throw in some actual numbers for "AO3 femslash is exploding":

Year	New works that year
2009	1266
2010	2444
2011	3677
2012	7466
2013	10977
2014	18972
2015	32308
2016	42055
2017	56958
2018	58281

(There are a zillion caveats to those numbers, but they're what was easy to gather, and the point is that the jump in the last five years has been huge. Does that reflect the fact that having a common platform with good tagging makes more space for smaller fandoms and exchanges, or does it reflect the fact that Supergirl happened? I cannot answer that question in the time I have available right now.)

Anyway, I'm happy to talk about this stuff more --- obviously I find it interesting. A lot of people are doing neat stuff, and I haven't done as much of a literature survey as I'd like to. It's a fun project, but I've also definitely found that it pays to not get too ambitious about what questions I want to answer (and not to be frustrated when my first answer to literally any random question someone asks about fandom stats is, "well, I... haven't the foggiest idea"), because answering individual questions is expensive.

[alias_sqbr]

Thank you very much for this comment, I kept leaving it until I had something interesting to say in reply but right now I don't. I find your posts really interesting, though.

[izilen]

My knowledge of maths and statistics is very poor, but your comment about historians being used to doing these sorts of approximations made me consider whether you had looked at any textbook-y sorts of resources aimed at, particularly, economic historians in training. Some of the methodology might be relevant? The other group of (academic) people I know of who regularly make estimations on the basis of very poor data with gaps and lack of information the further back you go in time are archaeologists, so there might be some insight there.

Regarding AO3 specifically: have you considered writing to AO3 Support? I know they have some boiler letters on the subject of statistics, but it is possible they might also be able to assist you with research on more specific things if you ask. They are very approachable people, and though they're generally buried in emails, they will always reply to each and every one.

I would also be willing to personally offer my assistance re: searching on the Archive. In all likelihood you know exactly what you're doing, but since I've started tag wrangling I've picked up on a lot of stuff on how to search the site etc and it might be that I could help you there. Just let me know.

[alias_sqbr]

Thank you! I don't have anything I want to poke at right now but I may hit you up later.

I just started a data science Coursera course run by health scientists, which looks like it'll cover this sort of thing. I used to work with health data (despite no very relevant training) and some of those data sets are very fuzzy.

[schneefink]

Fandom stats are fascinating :)

[alias_sqbr]

They are!