A study of Sherlock fandom

[petronia]

I've been incubating this idea since forever. I thought about it a lot, w/r/t AOS!Trek - not least because ayalesca was coming up with all these charts and things XD - but AOS!Trek was and is way too distributed, and contains too many characters and randomosity. (Never mind the super-gnarliness of Doctor Who.) BBC!Sherlock, however, is not only the perfect size, it's the perfect shape.** And although the main comm doesn't contain all the fic, it comes close enough. With 80% more effort, one can track the additional 20% at the old and new kink memes, plus AO3, but the kink memes feature massive reduplication versus the main LJ comm, and no content tags.

It's just a matter of paging back and filling in an Excel sheet, then doing scatter graphs. Frequency over time. XD I'm thinking:

* wordcount
* pairing(s)
* rating
* seme/uke, bawling this would require readthroughs but I feel like there wouldn't be any punch if I couldn't give the characters an indexed rating over time (even though, at the end of the day, my thumb-and-squint approximation will be borne out I bet)
* checkbox major categories, settings, and fandom tropes (here I'm interested in the way memes filter through a community. Asexuality, obviously, but also stuff like: there is proportionately way more animal transformation / animal familiar fic than I would have heuristically modelled. AOS!Trek contains tons of genderswitch)

** There's a major OTP and popular alternate pairings. There's extended!canon het and slash [EDIT -- and f/f, cannot forget], these-ppl-never-met crack, incest... the whole shebang, but no more than one or two combinatorial examples of each, because there aren't that many characters. XD; At the same time, no pairing is rare enough that there aren't a few stories, as long as ppl have bothered to come up with the idea to begin. And, of course, someone's posting fic every single day.

Comments

[ex_ayalesca554]

A+++++++

I LOOK FORWARD TO THIS. Sadly you are right, Trek will never be modelable XD;;;;;;

Am curious though - if you do this, do you think you might also throw in some kind of principal components analysis??? Would be v. interested in hearing what falls out of that XD

[petronia]

I don't really have the tools at hand for principal components analysis, but more to the point, what questions do you want to ask? XD

EDIT -- also, any advice on how to handle multi-chapter fic? The easiest would be to treat each chapter as its own story, but I can't decide whether that would be conceptually correct for stuff like pairing frequency count. o_O

[ex_ayalesca554]

Upon reconsideration, it's probably nothing that a nice regression wouldn't solve XD but I'd be curious to know what dimensions are redundant (although I would probably kick something if someone told me "statistically speaking, k/m and s/k stories are the same thing" ..) XD;;;; but that is more for hijinks than for production of real analysis.

Multi-chapter fic - do you mean like a series, or just a story released in spurts? IMO if the title changes, the new "chapter" should be counted as a new fic; if it's part n/m, then I would concatenate the chapter(s).

[troisroyaumes]

Yeah, I'd do a GLM and test hypotheses like length ~ pairing*rating or something.

Maybe multi-chapter vs. oneshot could be a categorical variable, and you could analyze those sets separately?

[petronia]

Like a WIP released in chapters. Even just a matter of answering the question, "how many stories containing pairing A were posted this week" if several of those were chapter updates... Sherlock fandom has a lot of ppl trickling out 30-chapter stories, which can be one chapter a day or one chapter a month.

[ex_ayalesca554]

Ahhhhhh. Well then I guess it depends on what *you* care about more, complete works vs. pieces, and make it clear what you mean. Although if you count "everything that belongs to the same story counts as one story", I don't know how you'd assign weights to WIPs, since if you don't know the final number you can't just assign it as 1/n ...

Tari's idea is also good -- you can run the numbers both ways, and see if it makes a difference *g* and report both.

[petronia]

Also, w/r/t Trek I would probably try to break it down into really simple questions and try to answer those individually... but you still run into all the sampling bias issues. XD; Whereas with Sherlock you can corral nearly the whole population.

[ex_ayalesca554]

The kink meme alone is beyond me I think XD; I WISH DELICIOUS ALLOWED FOR A NICE .CVS EXPORT DATA. Not that all things are even there -- though one could simply assume random sampling, which probably isn't too far from the truth.

[kickthebeat]

i will also be very interested in seeing what sorts of results you pull from such a study!