Actions

Work Header

[Fandom stats] Which are the biggest ships on AO3?

Work Text:

UPDATE: centrumlumina has done a more recent and detailed analysis on this topic. I recommend it if you're interested in this topic.

INTRODUCTION

Which are the biggest ‘ships in fandom right now? The Daily Dot posed this question, spurring my curiosity. I attempted to determine which ships had the most fanworks on AO3 at that time (August 2013).

Some important caveats and limitations to keep in mind:

  • this isn’t a good metric of how the size or intensity of a ship’s fanbase. There are various reasons I discuss below. This is only a good metric for the current state of what’s most common on AO3 in August 2013.
  • due to processing errors, about 5% of fandoms did not get properly analyzed - mostly ones whose names contain non-English characters, including Les Mis. This means the large ship Enjolras/Grantaire got left off the list — and possibly others. (Please help me spot errors and omissions!)
  • Merlin/Arthur (and possibly others) should be higher on the list, due to my using the wrong metatag for their ship.

RELATED WORK

Later, centrumlumina did some excellent work on the same topic, which I highly recommend (along with their other fandom stats work). (summary)

METHODOLOGY

I wrote a program that searched AO3 for every fandom, and grabbed the most popular 10 relationships for each fandom (as shown in the Sort & Filter sidebar when you look at search results), combined the lists, and sorted by popularity. For these purposes, I’m counting popularity as number of works on AO3 at this time; see Discussion for more thoughts on this metric.

For ships that showed up in multiple fandoms, I just selected the first fandom on the list. (Otherwise, the “Top 20” relationships would actually just be 9 relationships showing up in multiple fandoms) Removing duplicates didn’t have substantial impact on the ordering — e.g., if you just count the works in the Sherlock (TV) fandom, Sherlock Holmes/John Watson is still the most popular ship (18406 works).

My program did not properly handle fandoms with non-English characters in their names, and also omitted a big chunk of other fandoms, mostly in the "D" section of the alphabet. I hope to debug and repost an updated analysis in the future.

RESULTS AND DISCUSSION

[bigger]

As shown above, Johnlock is the biggest ship on AO3 by this metric, by quite a bit. Then Destiel and Sterek are nearly tied, with over twice as many fanworks each as most of the other ships. The raw data used to generate this chart is here (including one duplicate I failed to remove -- Loki/Thor):

[bigger]

Sherlock/John and Sherlock & John are both listed, because they are separate relationship tags on AO3 -- the second refers to friendship or other platonic relationship.

There are a some problems with using number of fanworks currently on AO3 as a measure of popularity. As I’ve noted in the past, the number of works posted on AO3 per month has shot up over the past years, and continues to do so, as shown in this graph created in May 2013:

[bigger]

That means that fandoms that are currently active (and their ships) are getting far more weight in this calculation than is truly representative of their popularity compared to older and currently less active ships. (E.g., it’s possibly that Spuffy was more popular in its heyday, relative to all the other ships that were being written about at the time, than Johnlock is currently. This analysis would not be able to capture that.)

I could try to partly address that issue by looking at the popularity of these ships over time and also trying to account for the different amounts of total fanworks produced on AO3 at different times (as I did when looking at Sherlock fandom shipping over time, for instance). I could divide the number of works produced about each ship by the number of works produced on AO3 over the same time range. Or, probably more sensibly, I could try to normalize the each work by the number of works posted on AO3 during that month — compare how popular each work was within the context of when it was posted. (So, for instance, if 10 works were posted in January, I could give each of the works posted in January a score of 0.1, and then I could sum up the total score generated by all works for each pairing.) That would give a more time-invariant picture of the popularity of each relationship, and I suspect the above graph would change substantially. However, that’s a much more difficult endeavor than what I did here, and I’m going to keep thinking about easier ways to normalize.

An additional limitation is that, some fandoms and/or some ships — especially the older ones — may have most of their works archived elsewhere, and might not have migrated to AO3. So again, comparing the sheer number of works on AO3 may be misleading.

Finally, the popularity of producing a type of fic is not necessarily the same as the popularity of consuming that fic, or the number of readers who will like the fic. More recently, I have more recently been exploring the differences between various popularity metrics and discussing their pros and cons. I may later revisit the matter of ships from some of these other angles.

In any case, I don’t think there is a perfect metric, here. But if you have other ideas for how to measure popularity of fandoms, ships, etc., I’d love to hear them!