YouTube's Content ID (C)ensorship Problem Illustrated
As we've pointed out repeatedly, poor design decisions in YouTube's "Content ID" system have resulted in over-blocking of videos that remix copyrighted materials. Today we got perhaps the most vivid example of the problem: the "silencing" of a lecture by Prof. Larry Lessig about the cultural importance of remix creativity. This is just the latest of many examples. We've been on YouTube's case for more than two years about this problem, and it's high time for YouTube to fix the Content ID system to respect the kinds of fair uses that are at the heart of remix creativity.
How did Prof. Lessig's video trigger the Content ID block? He included "snippets" (I use that word intentionally, as Google does in the context of its own Book Search product, to refer to small portions that should qualify as a fair use) from several remix videos. As a result, the audio track of his lecture included excerpts of several well-known songs. Apparently at least one of those songs is owned by Warner Music, which has chosen to automatically mute the audio track of any video when the Content ID system detects the presence of those songs. It's not clear which song triggered the block—the Content ID system doesn't tell you that.
Of course, in close cases, reasonable minds can differ about whether a particular use of a song qualifies as a fair use (although some cases are easy). But that's no excuse for the automated Content ID filter to block them—if a copyright owner has a good faith belief that any particular remix video crosses the line, it is free to send a formal DMCA takedown notice. Sending a notice is not hard, nor expensive, as demonstrated by the fact that copyright owners routinely send hundreds of thousands of these notices to YouTube. YouTube's Content ID system even will flag all the videos for the copyright owner's review.
But unlike the automated Content ID blocking, DMCA takedown notices at least put a human into the loop, and these humans must take fair use into account before issuing the notice. In contrast, an automated match by the Content ID system results in an automated removal, even where the copyright owner does not object to the use (and, as poorly behaved as Warner Music has been in the past, I can't imagine it really wants to censor Prof. Lessig's lecture).
Fortunately, YouTube permits users to "dispute" automated Content ID removals, and that's why Prof. Lessig's video is once again available. But that's not nearly good enough. First of all, YouTube's procedures for "removing" videos have created considerable confusion and consternation among users, and it's a fair bet that most YouTube users aren't aware of their ability to "dispute" these removals. Second, the thousands of lawsuits brought by record labels against individuals for file-sharing has created an atmosphere of fear that makes many YouTubers hesitant to go toe-to-toe with a major record label.
There's just no excuse for the Content ID system to be blocking remix videos. There's nothing in the law that requires YouTube to do this. In fact, section 512(m)(1) of the DMCA makes it clear that service providers do not need to install filters or monitor their services at all, much less allow copyright owners to use filters to block remixes.
Nor is there any engineering reason why the system should be designed this way. The filter can fix this problem by insisting that the audio and video tracks both come from the same copyrighted work and that the entire (or almost entire) video is drawn from the same copyrighted work. Unless these conditions are met, "block" should not be an option available to copyright owners. If a copyright owner wants to take down a remix video, they should have to follow the rules Congress established in the DMCA.
This is exactly what EFF, joined by numerous other public interest groups, asked YouTube to do in 2007 in our Fair Use Principles for User Generated Content. It's a shame that YouTube, a company that has become synonymous with remix creativity, can't find the time to fix its own Content ID system to protect remixers from unnecessary censorship.