rants

Is Crash the First of Many Search-Engine-Optimized Movies?

Why has the loathsome Crash clung to the top of Netflix’s top 100 chart like some preachy fungus? And why won’t it go away? Crazy bloggers are saying it has to do with quality or audience preference. We refuse to believe that this could possibly be true. Clearly, Crash has gamed Netflix’s search engine in four despicable (if unintentional) ways.

1. Timeliness.
Crash’s success is distorted due to historic accident: It won Best Picture in 2006, just as Netflix’s user base was exploding. It was likely one of the first films new users added to their queues. If Netflix had existed for twenty years, films made before 2004 might actually crack the top 10, thanks to long histories of user reviews and recommendations — but it hasn’t, and old movies haven’t.

2. Awards.
Despite the fact that Oscar ratings are plummeting and people seem to care less about award shows than ever before, awards seriously skew Netflix recommendations away from films people actually like and toward films that associations praise. Thanks to its confounding awards haul, Crash is linked to every other award-winning film in the database and sucks up referrals.

Could the problem with the vaunted Netflix recommendation engine be that it’s best at serving up search-engine-friendly films, instead of films we might really like? Do the number of name actors, number of awards, maudlin message, and timeliness steer users toward Crash instead of better movies? If so, maybe the competitors in the million-dollar Netflix Prize contest would improve their algorithms by penalizing films for having big casts, preachy messages, recent production dates, and truckloads of statues.

4. The Bradley Effect.
Clearly, it doesn’t hurt that Crash has a nearly inexplicable four-out-of-five star rating based on 2.8 million user reviews. The only rational explanation for such positive reviews is a film-crit Bradley Effect, a self-flattering bias in the star ratings that pairs nicely with Haggis’s middlebrow self-righteousness (so what if you’re a teensy bit racist, it’s not like racism is an institutional problem, and, according to Crash, you’re no more racist than the next guy and, best of all, there’s nothing to be done about it … ). Yes, the Bradley Effect was supposedly discredited in the last election — but can you really think of a more plausible explanation?

The implications of Crash’s Netflix dominance are scary. For decades, producers have been compiling crass and tidy charts of stars’ global worth (box-office numbers broken down by country and region). This is one of the reasons bad Brits and Aussies often get good American parts, regardless of their talent for dialects (see: Jude Law in Cold Mountain and All the King’s Men). Now that search engines are driving sales of video and movie tickets, will the casts get bigger and the roles smaller? Is this slew of ensemble films (we’re looking at you, He’s Just Not That Into You) a result of search-engine gaming? Is this why people keep casting 50 Cent, despite any evident talent? (See: the Crash-like ensemble drama Home of the Brave.) Is this why the similar, tastefully maudlin feel-bad message movie Babel is also perched at No. 13?

Could the problem with the vaunted Netflix recommendation engine be that it’s best at serving up search-engine-friendly films, instead of films we might really like? Do the number of name actors, number of awards, maudlin message, and timeliness steer users toward Crash instead of better movies? If so, maybe the competitors in the million-dollar Netflix Prize contest would improve their algorithms by penalizing films for having big casts, preachy messages, recent production dates, and truckloads of statues.

It’s worth a try. Crash must be dethroned.

Is Crash the First of Many Search-Engine-Optimized Movies?