YouTube Thinks "I Have a Dream" Is Lounge Music

For decades the "I Have a Dream" speech has been a rallying cry for civil rights everywhere. It's also been the centerpiece in a quiet, long and unfortunate copyright war, caught between those who argue it belongs in the public domain, where it can be accessed freely, and those who insist that it's in the private ownership of the King family and its foundation. ​In the process, videos of the speech have disappeared from YouTube, under a copyright controlled by the music publisher EMI, and a system that monitors copyright abuses on all of YouTube's videos.

In the latest chapter in the speech's copyright morass, that system, Content ID, has determined that at least some of the rights to the famous speech of August 28, 1963—and the profits it generates from each view—belong to a company called GR8 AL Music. That's because at some point since the video was uploaded last year, Content ID has decided that the video contains a lounge song titled "Our Dream," from an album called Lounge Cocktails, Vol.1 (Delicious Grooves for Café Bar and Hotel Suites), which itself contains a sample of King's speech.

In the past, YouTube's Content ID bots have marked uploaded videos of the speech the property of EMI publishing, and subsequently removed the audio or taken videos down entirely. Recently, however, Evan Greer, who uploaded the video of King's speech, noticed that it carried a peculiar ad.

"We were looking to share it to honor MLK's birthday and noticed that someone (not us) put an advertisement on it, due to YouTube's newish policy of letting artists put ads on other peoples videos if their Content ID system thinks their music is in your video," he wrote. "Given that many people click away from a video if it's got an ad on it, it sure feels like broken copyright has censored Dr. King's speech again."

The video was uploaded by Greer and the non-profit he works for, Fight for the Future, the copyright crusading group that grew out of the struggle over the 2011 Stop Online Piracy Act, SOPA. The upload itself was a form of protest, since unauthorized publication of the video violates its copyright. In the video, (Greer appended to the video one of King's famous quotes: "one has a moral responsibility to disobey unjust laws." Ironically, the letter from which the quote comes was later copyrighted by King himself.)

While the copy that Fight for the Future uploaded to Vimeo was quickly taken off the site, the YouTube version managed to survive. At some point, YouTube's copyright robocops confused the audio of King's speech with lounge music that contains a sample of the speech—a sample that itself is likely unauthorized.

First, remember the unlikely ownership of the recording of King's words. Though the famous oration was delivered on the Washington Mall to thousands of people, courts have ruled that its copyright belongs to the King family and to its rights protector, the London-based company EMI Music Publishing, the largest music publisher in the world. (And, because of a 2011 acquisition of EMI from Citigroup, the speech is now technically in the legal possession of Sony Music Publishing, the Estate of Michael Jackson, Jynwel Capital, the Blackstone Group, David Geffen, and the Mubadala Development Company, the government of Abu Dhabi's investment arm.) To show the speech in public, legally you must acquire a license, which you can get by purchasing a DVD of the speech for $20. Media outlets must pay higher fees; in August, CNN, ABC, and MSNBC aired the speech in full with the help of big sponsors like Bank of America.

Under these circumstances, the lounge song itself is likely violating the King copyright. But because of a fluke in the system (or two flukes), YouTube thinks that the composer of this lounge song—not Martin Luther King—is the original copyright holder, when in fact the musician simply copied and pasted King's words into an otherwise unremarkable hotel lounge soundtrack.

Why the King family's copyright hasn't been detected in the lounge song, and in turn, in the video of the speech, isn't clear. It's possible that both of these instances constitute "fair use," but no court has yet ruled on that question in relation to the King speech. In the case of the video, the fair use could be educational; in the case of the song, it could be artistic (even if it's not very good.)

In any case, YouTube's algorithms don't abide by any copyright law. The system in place for enforcing copyright doesn't follow the legal procedures of the DCMA, the law that relates to digital content. Instead, YouTube's copyright enforcement is based on YouTube's own terms.

Born out of a series of giant legal threats YouTube faced in its early years from the likes of Viacom, the Content ID system is bent in favor of copyright holders, not creators or artists. Under YouTube's terms of service, there's no penalty for making an erroneous claim, which creates little incentive for precision, and perhaps an inclination for some amount of corruption.

There are other problems too. Countering a claim doesn't automatically remove the claim; until the matter is resolved—a process that can take weeks—the claimant continues to draw profits from ads on the video. Meanwhile, YouTube's take—45% of all profits—are the same, no matter whether EMI or Adam the Alien claims the rights to content, giving the site little financial incentive to correct flaws. In late 2012, YouTube issued reforms to the appeals process that were intended to make it easier for YouTubers to counter erroneous claims, but the changes didn't address more basic problems, like intitial fraudulant claims or the mistakes of algorithms.

Over Christmas, a YouTuber named Adam the Alien was the subject of another strange copyright charge: a number of multinational companies claimed they owned rights to Adam's rendition of "Silent Night." Never mind that the song was composed in 1818 and was firmly in the public domain. Once their claims had been made under the Content ID system—either accidentally or manually—all the money that Adam's video generated was diverted to them.

"As an independent content creator, it is absurd, ridiculous, and downright insulting that I can have my content de-monetized based on a completely fraudulent claim," he wrote on his blog. "The fact that the claims are based on an automated system doesn’t make it any better. If anything, it makes me think the automated system should not be in place. Or at the very least, it needs a major overhaul, and a lot more human eyes involved before action is taken."

In December, Content ID ignited anger throughout one of the largest segments of YouTube's base. A week before Adam received his claim, thousands of video game vloggers found themselves the target of automatic claims by dozens of video game companies. Their crime: posting popular "Let's Play" videos to their channels—walkthroughs, playthroughs, and reviews—just as they'd been doing for years, in ways that the gaming companies had long considered a matter of "fair use." As it stood, the system helped to promote the companies' games while providing a small income to thousands of bloggers. The backlash was intense, prompting many video game companies to publicaly decry it and help bloggers dismiss the claims.

Last year, when Harvard law professor and copyright reformist Lawrence Lessig was targeted by Content ID for including a bit of a song by the band Phoenix in video of a lecture of his that he uploaded to YouTube, he was even warned that he could lose his account. After an unsucessful counter-claim (his was a "fair use") and the threat of a lawsuit by the claimant, Australia's Liberation Music, Lessig decided to take Liberation to court for acting in bad faith. He seeks unspecified damages from the company, including legal fees.

The false positives can get even stranger. In some cases, YouTubers have been punished for posting videos widely identified as fair use or public domain, simply because another organization (falsely) claimed the content as their own.

That's what happened last year, when NASA uploaded its own footage of the Mars rover landing to its YouTube channel. Within minutes, the video was flagged for infringing upon the copyright of the media company Scripps. Apparently, someone at Scripps had managed to upload a copy of the publically-owned footage to Scripps' YouTube channel around the same time as NASA had, but they had failed to indicate that the video was in the public domain.

As a result, the video wasn't just flagged: because of Scripps pre-established copyright claims, the video was automatically pulled off YouTube. For a few hours, those looking for a video of the rover landing would not have found the NASA video; instead, they might have seen a version with ads on it, with revenues going to Scripps. "Ironically," writes gaming blogger Mona Ibrahim, Content ID "facilitates a system whereby the original content owner becomes an infringer."

There are lessons for all sorts of algorithmic bad behavior here, not least of which should for be the systems used by the NSA today for picking through haystacks of metadata for needles. Data does lie, and algorithms aren't always right. If you think the stakes are high when algorithms are applied to copyright law, consider what they could mean in the context of surveillance. (The pursuit of MLK by the NSA and the FBI offers other lessons about the risks of a domestic spying machine.)

The immediate lessons may be for the people who administer systems like Content ID. The Content ID system is necessary, YouTube argues—there are over a hundred hours of video uploaded to its servers every second. But it's also flawed, relying upon a “guilty until proven innocent” doctrine, one that automatically diverts ad revenues into the pockets of claimants, whose claims can only be reversed if the claim is disputed, a process that can take weeks.

Watching the accidents of our copyright robocops crop up increasingly often is like listening to a bad lounge song on repeat. Until we find a better way to discuss and enforce copyright online, the "fair use" of creators and the legacies of figures like Dr. King will be tied up in the intricacies of digital copyright management. That sets up roadblocks for distribution, and sometimes confuses one of history's most famous speeches for a piece of Musak.

@pasternack

Read more on the "I Have a Dream" debacle here. You can read and listen to the “Dream” speech and other of King’s speeches at American Rhetoric, and can make a donation to the King Center here. More on online copyright:

YouTube's New Copyright Bot Pwns Gamers

Copycats, Takedowns, and Ass Rainbows: What Does Copyright Mean for Internet Memes?