Form v. content in DMCA notices

Still working on a long post on Garcia v. Google because ugh, but here I go on a side note: David Post has a post up, Why Google shouldn’t be the copyright court of last resort, which argues that Google shouldn’t screen for bogus copyright notices. While I agree with the title of the post, I’m not persuaded by the content. Post says Google should’ve just honored Cindy Garcia’s takedown request, because it had the proper form of a DMCA request. He argues that the result of Google’s refusal was a case in which the defendant, Google, didn’t have any access to the true facts, making for a bad contest on the question of infringement.

I think Post’s proposed solution—honor all takedowns that are properly formatted—is a pretty bad idea, despite the terrible opinion in Garcia.  (Also, what’s up with “last resort”?  As the 9th Circuit opinion demonstrates, Google is pretty clearly not the last resort, but rather the first screener.)  Post says that the proper response here was to wait for a counternotification, and the posters themselves could’ve fought about the underlying facts.  Except … I personally know plenty of people with valid fair use arguments who’ve decided not to take the very tiny but intimidating to nonlawyers risk of counternotifying; the research suggests that counternotification is vanishingly unlikely, even when (as is not uncommon) the notices appear to be invalid after minimal scrutiny; and the video here seems to have been reposted by many people who weren’t the copyright owner either in order to make some sort of point about Innocence of Muslims, making it even less likely that we’d get the “right” result through counternotification.

Percentagewise, most notices are valid—but, as Post points out, this is a mostly automated process; if even 1% of 100 million notices are invalid, the absolute number of bad claims is very high, something a free speech-sensitive analysis ought to be concerned with.  Automatic compliance with everything that looks like a DMCA notice would readily enable low-risk censorship and vitiate §230, which protects against such demands when the offensive subject matter isn’t infringing.  This isn’t hypothetical.  People send DMCA notices when they object to use of their trademarks (also here) or have other noncopyright claims, and I heard trademark counsel advising in favor of using copyright to enforce trademark claims at INTA.

When DMCA takedowns target obvious fair uses, are based on non-copyright claims, or are based on politics, a service provider should be able to reject them.  Here’s some other randomly selected examples of DMCA failure, out of many: Warner Brothers sending DMCA notices for programs it didn’t like but didn’t happen to own copyright in (mistakes attributed by Warner to computers); DMCA notices sent to film critics; DMCA notices sent to take down political ads; Larry Lessig’s first DMCA fight; Larry Lessig’s second DMCA fight; various anticompetitive/abusive takedown notices to WordPress; Apple attempting to suppress public circulation of its contract terms; suppression of the Hugo Awards show for including licensed content; etc.

Google is mitigating some of the damage by screening some notices that are problematic on their face (as Garcia’s reasonably could be seen to be, since she doesn’t even claim a copyright in the film and since the film was controversial for other reasons, making the “copyright as censorship” problem a real risk).  So, by the way, is Wikipedia, which also has a DMCA policy but does not automatically take down content without independent review.

Google shouldn’t have to be a copyright court. But until people stop sending bogus DMCA notices, perhaps because they are required to stop doing so by more robust §512(f) enforcement, it’s better than the alternative.