Inspired by billiethepoet's recent analysis of the holmestice fic exchange, I did a purely descriptive statistical analysis of the contributions to acd_holmesfest over the years. Basically I wanted to put some numbers on stuff I've been curious about for ages.
This was done by scraping the comm pages with the python libraries requests and beautifulsoup, and then cleaning and analysing the data with standard python. The acd_holmesfest mods have been very consistent in tagging and titling entries and getting people to use headers, so it was possible to automate the process almost completely. I'm not a statistician, and also I've been a bit lazy in terms of significance testing and so on. If you spot any mistakes (like, your 100,000-word epic fic is clearly missing...) then please tell me!
It's interesting to compare to other Holmes-related exchanges, notably holmestice (an exchange for all Holmesian adaptations, dominated by BBC Sherlock) and to fandom in general, though we're obviously working with a much smaller population here. I do that at various points, where I've managed to find some other stats to compare to. (Other suggestions welcome!)
There have been six rounds of acd_holmesfest. Let's start with the number of works per round (fic, art and podfic combined; illustrated fic is counted twice, as is fic written specially to be podficced for the exchange).
There have been some rather large fluctuations, but overall the number of participants has held pretty steady since the start.
Then I went on to look at things like rating, medium etc. The rest of the figures are summed over all 6 rounds, since in each case the numbers were fairly constant as a function of time.
First of all the type of work: fic, art, or podfic. Again, illustrated fics were counted twice, once as fic and once as art. Fic written and recorded specially for the exchange was also counted twice, once as fic and once as podfic.
Most participants wrote fic, but personally I've been happy to see podfic in almost every round :)
Now let's concentrate just on the fic, for attributes that don't exist or aren't always defined for art.
Fics are all rated according to a scheme suggested in the mod-provided header (G, PG, R or NC17, i.e. something very similar to the MPAA film rating scheme).
There's a clear dominance of PG-rated fics. This seems to be part of a general trend across fandom. For example in the holmestice fic exchange, the T rating (more or less equivalent to PG) is by far the most popular. See also this survey of the fic preferences of AO3 users in general, where T is the rating people most prefer to write (though the numbers there aren't the answer to the question 'How many works fall into each category', so aren't really comparable to the numbers here).
Now for the distribution of fic lengths. You should read this figure as the number of fics whose wordcount fell in the range 0-1999 words, 2000-3999 words, etc.
acd_holmesfest has no minimum (or maximum) word limit, but there's a clear preference for fics in the range 2000-4000 words. ETA: Actually, speaking more precisely the peak is at 2000-3000 words. The longest fic was around 14800 words!
Next up, the breakdown into 'gen' fic (which don't focus on a romantic pairing) and 'not gen' fic (which do focus on a romantic pairing, whatever that may be). I defined gen as any fic that didn't contain the pattern 'character name/character name' in that part of the header which gives info about characters and/or pairings. The use of the '/' to denote a romantic pairing seems pretty standard in this part of fandom. The 'unknown' is for a small number of fics where I couldn't find any info about pairings or lack thereof in the header.
There's been slightly less gen than not-gen, though in one round (round 3) the gen actually outnumbered the not-gen.
This is another point where it's interesting to compare to other stats. Destinationtoast has an analysis of fics on AO3 tagged as ACD canon as of June 2013, and it looks like about 35% gen (though the real number of purely gen fics is probably lower, since that number includes fics that are tagged as 'gen' and something else at the same time). The holmestice fic exchange has been 25% gen over the years (remember this one is for all versions of Sherlock Holmes, dominated by BBC Sherlock). AO3 users in general report producing 23% gen (though again, this last statistic, unlike the others, is not the answer to the question 'How many works fell into the 'gen' category and therefore isn't really comparable).
The not-gen fics can be further broken down according to what exactly the pairing is.
The most popular pairing by far is Holmes/Watson, unsurprisingly, and the next most popular is Watson/Morstan, though there are actually less fics in this latter character than in the category of 'all other pairings'.
Instead of looking at which exact characters are in a pairing, we can make broad categories:
- slash, or M/M pairings, so including Holmes/Watson, but also such pairings as Watson/Lestrade or Holmes/OMC
- het, or M/F, which is basically Watson/Morstan and one each of Holmes/Adler and Adler/Norton
- and finally threesome, of which there was only one (Watson/Morstan/Holmes)
Well, that's about all I could think of looking at... If anyone wants the raw data for something, let me know!