I’ve been an anime fan for a decade now. I’ve watched a ton of anime and dabbled in different fandoms. As outstanding members of the anime online community, we may remember that moment in time when our anime passion settled down for the first time. We started noticing repetitive patterns in storytelling and character design. We wondered whether the novelty effect of a product from an exotic land finally dissipated. We thought to ourselves whether we have outgrown the medium we loved and staked our reputation on even a few months ago.
This is the point where anime fans either stay fans for a prolonged period of time or ditch anime completely, cursing on their way out what a colossal waste of time that was. The ones that stay slowly develop an appreciation for these repetitive patterns and elements. We develop our tastes, we realize what we truly want from anime, what genres and themes interests us. In short, we learn what turns us on. Some fans feel very strongly what story tropes and character traits excite them, they even go as far as to catalogue them on sites like TV Tropes. Not my kind of fun, but I do see tangible worth in such crowdsourcing efforts.
Say you are an anime producer. You have a general idea what the fans want, but at the end of the day you have to be sure to a certain degree. No doubt, with your next planned anime production there’s money to be lost. You don’t want your title to operate at a loss, as most television anime do, so you’ll be sure to follow the appropriate fandoms, read every product feedback you received from your customers and look at the latest trends from Comic Market before you begin pre-production.
This past holiday break I began working on a project with a very lofty goal—to make a computer model that will be used for predicting future anime winners, and it will be analyzing tropes to do just that! Sounds ridiculous, right? The theory goes tropes—or combinations of them—aren’t just objective meta information for a certain title, they carry a fair amount of subjective sentiment that we may interpret as hidden knowledge still waiting to be discovered. Say we take all those tropes, associate them with user-submitted scores for those particular anime titles and make a model of it. If and only if we’re descriptive enough with the tropes we specify, we may be able to predict new content winners with good enough accuracy.
Of course, in anime production execution is king, but that’s a different subject for a different person.
Right now, I’ve finished part of the first phase of the project, which is data collection, and decided to share my findings so far, to keep my motivation high. The post image is a word cloud which consists of the two hundred most common tropes posted on TV Tropes for television anime for the 2011–2014 period. Visualizing the information was nothing special in terms of work. Most of the effort went into writing scripts for scraping the site and organizing the information collected. Let me just say their wiki is pretty messy for doing that sort of thing. If anyone is interested in the code, hit me up on Twitter.
So, which tropes are the most common? Here’s the list:
- Wham Episode
- Gratuitous English
- Beach Episode
- Running Gag
- Mood Whiplash
- Curb-stomp Battle
- Meaningful Name
Let me just say that Shoutout is by far the most common trope, so I’ll probably have to treat it as an outlier or as a non-informative feature. Commentary about the rest of ‘em? As expected. As an anime producer, make sure that the show has at least one shocking episode, a tsundere girl, and lots of fanservice. By then you can be sure you have a winner on your hands. Well, it’s not that simple, as we will see over the following months, but it makes for a good joke.
I later found out (as I’m pretty new to trope wikis) that TV Tropes may not be the best trope wiki out there. As it turns out, the site has banned a few series due to arbitrary censorship, for example Ro-kyu-bu! I’ve decided to redo my trope collection phase at a later time on a forked site called All The Tropes, or maybe I’ll use it as a secondary source. I also want to include Japanese tropes and tags, so if anyone happens to know Japanese sites with such information available, I’ll be eternally grateful for the hint.
Next time I’ll be looking over anime grades submitted by users of popular database sites such as Anime News Network and MyAnimeList.