YouTomb Takes Stock Of YouTube Takedowns
Site has Public Database of Video Removals
On April 6, 2008, a clip of an Oklahoma police officer assaulting a man appeared on the popular video-sharing site YouTube. Tagged with “police,” “brutality” and “beat up,” it received over 20,000 views in the ten brief days that it was available.
Taken from inside the police cruiser, the clip shows a tan Toyota pickup stopped along an icy, tree-lined road. The officer is telling the man in the passenger-side seat to step outside. “Sir, turn around and put your hands behind your back,” the officer says, moving to make an arrest.
The man opens the door, but holds up his hands. “Whoa, wait, no …” he says. He is confused; this must be a misunderstanding.
Again the officer barks his order, but the man hugs close to the truck, wary: “No, wait quit it! What’s going on?” he asks.
At this point the officer grabs the man and sprays his face with pepper spray. The man screams and screams.
The rest of the video is hard to understand because the yelling has overloaded the microphone.
In response to this video, the city of Blackwell, Oklahoma promptly filed a copyright infringement complaint. On April 17 the clip went down, joining an ever-growing graveyard of videos removed from YouTube — some for good reasons, some for bad, and some, seemingly, for no reason at all.
To Gregory Price G, David Sheets ’09, Quentin Smith ’10 (also a Tech staff reporter), Dean Jansen, and the students of MIT Free Culture, the removal of the Oklahoma arrest clip was yet another example of copyright law abuse on YouTube, and exactly why they created the YouTube monitoring site YouTomb last year.
YouTomb, available at http://youtomb.mit.edu, is a record of YouTube videos lost. On YouTube, barely a trace remains when a video goes down. For the police brutality clip, which Price discovered using the site, there is only this message: “This video is no longer available due to a copyright claim by City of Blackwell, Oklahoma.”
YouTomb gives context to the takedowns: it doesn’t save the video file itself, but it records the video’s description, when it was uploaded, how many views it got, thumbnails — details that might explain how and why a video was removed. Over time, patterns found in the data could show how copyright is enforced on YouTube and might help devise ways to protect legitimate videos.
So far the project has recorded over 43 gigabytes worth of data covering a quarter of a million of YouTube’s most popular videos. The database, available online for anyone to access, has already yielded interesting finds like the Oklahoma arrest clip, but it’s not just for monitoring copyright abuses any more — it now keeps tabs on international censorship as well.
Copyright confusion
Copyright problems have dogged YouTube ever since users realized that a site designed for sharing home movies with friends could also be used to share Hollywood movies with the whole Web.
The Digital Millennium Copyright Act protects sites like YouTube from getting sued for hosting these copyrighted video, but only if YouTube promptly removes them upon request.
“YouTube takes the position of neutrals,” said Corynne McSherry, staff attorney at the Electronic Frontier Foundation. “They say, ‘If we get a [DMCA] notice to take a video down, we’ll take it down. If we get a notice to put it back up, we’ll put it back up.”
According to the students of MIT Free Culture, a student group promoting free speech, open software, and fair use on the internet, YouTube’s policies are unfair toward users, partly because the DMCA itself favors copyright holders.
“Under the DMCA it’s incredibly easy to get stuff taken down, no question about it,” McSherry said.
“Back before the DMCA, 1998, if you wanted to get material down off a Web site, you had to go to court, you had to get a restraining order,” McSherry said. “Nowadays, you just send an e-mail and [Web site operators] have to take it down within a certain time.”
But the video-sharing site has made the process one step easier: it offers a convenient web-based form to issue DMCA requests. And, according to members of Free Culture, YouTube does not immediately check to see if the requests are real or fake.
This open system has attracted abuse. According to the EFF, over 4000 videos criticizing the Church of Scientology were taken down on Sept. 4–5 because fake DMCA copyright infringement notices were filed by American Rights Counsel LLC. The group does not appear to have any connection to the videos whose copyright it claims to own, some of which are news clips from foreign television channels. Many critics allege that Scientology is a dangerous cult notable for using any means necessary to silence critics.
Censorship also seemed to be the motivation for the takedown of the Oklahoma police brutality video, which Price found while casually surfing YouTomb one day.
It’s not clear if the Blackwell, Oklahoma city government owns the copyrights to the video, as works created by government agencies are generally not protected by copyright. But, copyrighted or not, Price said he thinks the real reason for the video’s removal was to save the police department from embarrassment.
On the YouTomb blog, Price wrote: “[i]f anything qualifies for fair-use protection, it’s a citizen showing their neighbors what their government is up to — the core of the First Amendment’s guarantee of free speech. Even if the city had the law behind it, it wouldn’t change the motive of this takedown: censorship.”
Future fair use
The idea of fair use is dear to Price, who helped start the Free Culture club at Harvard when he was an undergraduate, and founded MIT’s Free Culture club last year as a graduate student in Course 6.
Under a provision of U.S. copyright law called “fair use,” the free use of copyrighted works is sometimes allowed. People who copy materials for “purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research” are relying on the U.S. code’s protection. Fair use lets MIT professors post scanned copies of books onto Stellar sites and protects people who make parodies from getting sued.
But in modern law, the definition of fair use is muddy. MIT Free Culture created YouTomb in part to monitor YouTube for improper takedowns of videos that should be protected by fair use but are swept up with the rest of the infringing videos containing copyrighted material.
“A lot of the [YouTube] takedowns are legitimate,” Sheets said. “It’s the ones that aren’t legitimate that are important.”
Around the time that the project began, YouTube also started implementing a video ID system that automatically scans uploaded videos to compare them with copyrighted content. The idea of a machine enforcing copyright law especially disturbed the volunteers on the YouTomb project.
“[Identifying cases of] fair use requires a human,” said Oliver Day, a researcher at the Berkman Center for Internet and Society at Harvard who helped build YouTomb. “Even trained lawyers debate over what is and is not fair use; a program can’t do it.”
“The fair use exemptions are intentionally quite vague,” Price said. “There’s no time where you can definitively point to something and say ‘That’s fair use.’ Even if you’re fairly certain it’s fair use, it would still take forever in court and cost a lot to defend.”
One such debate that was recently settled in court involved a YouTube video Stephanie Lenz uploaded of her toddler dancing to Prince’s song “Let’s Go Crazy.” Universal Music Group demanded that the video be taken down. Lenz not only got her video put back up, but sued Universal for trying to enforce copyrights on a work that was fair use.
In August, the judge ruled that copyright holders like Universal must think before it sends a takedown notice, no matter how easy the DMCA and YouTube make the process. In order to send a takedown notice, the judge ruled, a copyright holder must reasonably believe that something is infringing and is not protected by fair use.
That ruling alone will not protect free speech on YouTube, Price said. Many users may be confused or intimidated by the takedown process, and few are likely to have the support of the EFF, as Lenz did. Few know that even if their video is taken down, under a provision of the DMCA, they can challenge the notice and have their video put back online.
The process involves filing what’s called a DMCA counter-notification, which is document attesting that the content taken down was not infringing on anyone’s copyright. YouTomb does not compile statistics on the percentage of people who file counter-notices. But McSherry thinks many don’t take advantage of this process.
“If you get a notice that one of your videos is taken down because you broke the law, that’s kind of terrifying, and people just don’t know what their rights are,” McSherry said.
‘Terms of Service violation?’
As complicated as a copyright infringement suit can be, sometimes videos are taken down or censored and there is no clear recourse.
In addition to tracking videos taken down for violating copyright law, YouTomb also tracks videos removed for violating YouTube’s own Terms of Service. Many of these videos are pornographic, but some seem innocent: A video of green lasers lighting matches taken down on Sept. 8; or a video of a tornado destroying a house taken down on June 18.
The YouTube Terms of Service are an area of particular concern for McSherry, she said, because the takedowns can be arbitrary and unexplained.
“In the Terms of Service, there isn’t any kind of formal protection, or a formal procedure for users to get their data back except to send a letter to YouTube begging them to reconsider,” McSherry said.
YouTube did not respond to requests to comment about their terms of service. In a statement, a YouTube spokeswoman said that “Our community polices the site for inappropriate material, and users flag content that they feel is inappropriate. Once flagged, content is reviewed by our staff and usually removed from the system within minutes if it violates our Community Guidelines. For more information, please visit: http://www.youtube.com/t/community_guidelines.”
No NBA in China?
Day has been monitoring another type of censorship on YouTube — country blocking. Some videos on are inaccessible in certain countries, which the MIT Free Culture team found out this year when they discovered that many NBA videos were specifically blocked in China.
They brought the data to Day, who has worked on censorship in China. He wrote scripts to search the YouTomb database and found many more examples of country-specific censorship. France and Germany, for instance, block Nazi and hate-speech related videos, though these videos are viewable from other countries. In Thailand, videos critical of the king are blocked. The BBC has its own channel on YouTube that is blocked in every country except the UK. The Chinese government’s Olympics YouTube feed was unavailable within the United States, where NBC had an exclusive broadcast deal.
Smith said it’s unclear if these blocks are made by governments, YouTube or content creators, though he guesses that any of the three could be involved.
“Right now we’re sort of in the collecting data phase,” Smith said.
Indeed, much of the project’s data, while publicly available, has not been thoroughly analyzed.
“There have got to be interesting things in [the database]. Right now we just need someone to be like ‘I’m going to datamine the hell out of this,’” Sheets said.