Posted on Thursday, January 2nd, 2014 by Angie Han
Among the Netflix faithful, the service’s hyper-specificity is something of an affectionate joke. Just about any movie database will differentiate between, say, drama and comedy, but Netflix comes up with categories like “Critically-acclaimed Witty Independent Comedies” or “Imaginative Time Travel Movies from the 1980s.”
Just as you’d imagine, those microgenres add up — to 76,897 unique descriptors, as counted by one journalist. And they reveal some interesting information about what exactly we’re all watching on Netflix. Hit the jump to see what he found.
The Atlantic’s Alexis Madrigal got curious enough about Netflix’s many microgenres to write a script compiling all of them. It took the program 20 hours to find 76,897 different categories, though it should be noted that some of the categories are empty (i.e., don’t have any streaming titles associated with them).
At the base of each microgenre is a broader traditional genre, like drama or comedy. From there, Netflix mixes in a whole bunch of other filters — such as location, time period, actors, and more — to get to their super-specific categories.
Netflix is famous for using complicated algorithms to cater their recommendations to each individual user’s (stated or unstated) preferences, but the process isn’t all mechanical. First, films are analyzed by an army of (human) taggers who rate the film across various criteria. Their values are then put through the Netflix process.
After poring over thousands of niche categories, Madrigal found that certain patterns started to emerge. For example, while these microgenres covered all kinds of movies, certain themes were more popular than others. For example, microgenres about marriage and royalty were among the most common, while those about September 11 were among the least common.
He also discovered that “Romantic,” “classic,” “dark,” and “critically acclaimed” were some of Netflix’s favorite adjectives for movies. The 1980s were the most popular time period, followed by the ’70s and then the ’60s. And “set in Europe” was by far most common description where setting was described at all, presumably because Netflix doesn’t bother telling Americans if something is set in modern-day America.
All of which is interesting, but not especially surprising. What was more unexpected was Madrigal’s result for the actor with the most microgenres to their name. Beating out Bruce Willis (No. 2) and Robert De Niro (No. 6) for the top spot was Raymond Burr, star of the ’50s series Perry Mason. Not even Todd Yellin, Netflix’s VP of Product Innovation and the guy who came up with Netflix’s crazy system, could explain that one.
For much more on Netflix’s tens of thousands of microgenres, including some really interesting details on how the system works, click over to The Atlantic to read the full article.