Why you shouldn't blame YouTube for removing all of those Hitler videos.

Why you shouldn't blame YouTube for removing all of those Hitler videos.

Why you shouldn't blame YouTube for removing all of those Hitler videos.

Innovation, the Internet, gadgets, and more.
April 27 2010 5:29 PM

YouTube vs. Der Führer

Why you shouldn't blame the video site for removing all of those Hitler videos.


A little more than a week ago, just before Hitler's birthday, Constantin Film sent YouTube the "reference files" for Downfall, the 2004 German movie about the dictator's last days. Constantin, which owns the rights to Downfall, asked YouTube to find and take down every video that included a clip from the film. YouTube can do this automatically—it plugs a studio's reference files into Content ID, its copyright-infringement search tool, and then mines its enormous video database for any clips that match the files. Over the course of a few days last week, Content ID turned up lots of videos that appeared to infringe on Downfall. The system blocked them all.

But Content ID is a soulless computer program; it doesn't understand sharp political satire, trenchant sports commentary, or biting cultural criticism. Content ID thus couldn't tell that it was pulling down videos that were likely protected by the fair use exception to copyright law. Many of the Downfall clips on YouTube were inspired by one of the funniest Web memes around—videos showing Hitler fuming about the war paired with English subtitles that reveal him to be angry about something altogether less serious. Among the videos Content ID removed were those of Hitler ranting about getting banned from Xbox Live, Hitler finding out about Michael Jackson's death, Hitler as Hillary Clinton reacting to losing to Barack Obama, and the Funny or Die clip embedded below—Hitler finding out that bloggers remember him mostly as an Internet meme ("I slaughtered millions, cut a bloody path of destruction across Europe! And for what? So I could be the latest juvenile Web fad?")


There's been a lot of outrage online about the clips' disappearance from YouTube. The company's critics say the Downfall takedowns bolster their longtime complaints about Content ID—that the system gives too much power to studios, letting them censor videos that are clearly legal and giving uploaders little recourse to fight back. In the past, Content ID has pulled down, among other things, videos showing kids singing copyrighted songs and a lecture by law professor Lawrence Lessig about the cultural importance of Internet remixes. Both the director and writer/producer of Downfall have said they love the parodies. Shouldn't YouTube have known that these clips weren't infringing? Couldn't it have applied a little common sense?

For YouTube, common sense isn't really an option. A site that sees about 24 hours of new video uploaded every minute could not exist if it had to adjudicate copyright disputes manually. A few weeks ago I criticized YouTube for its founding sins; documents uncovered in the billion-dollar lawsuit that Viacom filed against YouTube suggest that, in order to keep traffic booming in the site's early days, YouTube's founders decided to ignore copyright concerns. Content ID, which YouTube began building after Google purchased the site in 2006, is a brilliant response to those early missteps.

The fights between content owners and people who remix their movies and songs have been among the most contentious issues of the digital age. Content ID is a typically Googly way to solve these disputes—it replaces human judgment with powerful algorithms. The system is not perfect, but it's not as terrible as its critics insist. Don't blame YouTube or Content ID in the Downfall flap; this was entirely Constantin Film's fault, and we should all send our dictatorial rants its way.

At its heart, Content ID is like a souped-up version of the FBI's fingerprinting database. The entertainment industry keeps sending YouTube new reference files for movies, TV shows, songs, video games, and other content. YouTube scans every new upload and the millions of videos in its database against each of these files. David King, a YouTube product manager, told me that the system can find extremely fuzzy matches. It can spot when a copyrighted video has been transformed in some way by an uploader—for instance, it can finger a basketball game even if you pause, rewind, and then replay a clip from it, and it can identify Eric Cartman if you record a clip of South Park by holding your camera up to your TV.

But clip-matching is the easy part of Content ID. The hard part is deciding the legal and business implications of each attempted upload of copyrighted content. That's what's so clever about Content ID: The system allows studios to leave user-uploaded copyrighted clips on the site. Instead of taking videos down, they can choose to run ads against them; YouTube has worked out a series of complex deals with studios and record labels to automatically share revenue from these ads with content producers. YouTube says that a third of the money it passes out to content producers is generated by Content ID matches—in other words, tens of millions of dollars or more are generated by videos that were uploaded by YouTube's users, not the industry itself. And these user videos can generate far more cash than industry-approved uploads. Remember that video of Jill Peterson and Kevin Heinz's dance-music wedding march? * It's been viewed more than 48 million times. Content ID flagged its soundtrack as containing "Forever" by Chris Brown, and Sony Music ran ads pointing users to sales pages on iTunes and Amazon. The song rocketed up the charts.