Doing the job, but not a very good job
When people talk about effective ways to mitigate the impact of online piracy, YouTube’s Content ID is often used as an example of what works. Unfortunately, despite its role as poster boy for anti-piracy tech, in reality it falls flat as a gatekeeper against online piracy.
Aside from a labyrinth-like user interface that seems likely to have been designed–not to help– but to discourage rights holders from using Content ID, the actual fingerprinting technology behind it can be easily fooled.
YouTube introduced the Content ID system in 2007. At the time, the company was facing pressure from a Viacom lawsuit, among others. According to YouTube, it’s pretty straightforward:
Videos uploaded to YouTube are scanned against a database of files that have been submitted to us by content owners. Copyright owners get to decide what happens when content in a video on YouTube matches a work they own. When this happens, the video gets a Content ID claim.
Looking to make money off work they don’t own, clever YouTube users have discovered ways to fool the technology so their illegal uploads of copyrighted movies and music don’t get flagged, blocked or removed.
I began noticing this phenomenon more lately as I’ve begun to find full, infringing copies of films uploaded that matched content owned by a film distributor I work for. This seems to be happening more often and I was curious as to how these pirated copies had avoided detected by Content ID. When I looked closely I saw that subtle manipulations in brightness had taken place along with slight adjustments to frame size and sometimes the crop of the frame.
When I started poking around YouTube to find other examples of these uploads they were easy to find. It only took me a few minutes to find dozens of copies of a variety of full copyrighted movies, old and new. One title I came across was the movie, Everest. Below are screen captures from two different full uploads of the movie I found streaming on YouTube.
Two full copies of the movie Everest uploaded to YouTube.
In this case the uploader had used several techniques to avoid detection including reversing the frame (note the backwards title), darkening the lower part of the frame and cropping it. Of course, having recently viewed the film on HBO, watching a lousy copy like this on YouTube wouldn’t be my choice, but apparently others didn’t mind. Uploaded only a month ago, the movie had already racked up more than 16,000 views.
Pirate uploads make money for uploader and for YouTube
Why go to all this trouble to manipulate a movie for upload to YouTube? Well, it’s the age-old pirate motivator–money. This uploader, who goes by the name Kenneth Lamb, has claimed ownership of this content and monetized it with ads. He makes money. YouTube makes money. The movie’s actual production companies make nothing.
This YouTube user claims to own rights to Everest movie worldwide and makes money off ads
In an ironic twist, several of the ads that appeared when I was examining (and reloading) this pirated copy of the film were for films including DreamWork’s upcoming movie Trolls and Warner Brother’s Jason Bourne. It’s more than a tad ironic that Hollywood studios are (inadvertently) putting cash in YouTube’s hands via advertising on a pirated copy for one of its own productions.
Ultimate irony that ads for upcoming movie releases are featured on pirated copies of Hollywood films
I don’t deal with music or audio files on YouTube but there are similar manipulations happening there as well where uploaders resample, add noise, etc. to fool the Content ID system into ignoring the file.
What can YouTube do to fix this growing problem? Per usual, the list is long and varied, but begins with asking Google engineers to design better fingerprinting tech. There are other companies that offer digital fingerprinting technology seem to do a better job catching these circumventions. If I can easily uncover an upload is a copy of the movie Everest, why can’t Content ID? You can’t tell me that with all its financial (and technological) resources YouTube doesn’t means to upgrade its system?
Technological solutions exist. It’s just a matter of priorities. Stopping piracy isn’t a priority for YouTube.
Aside from updating its fingerprinting capabilities, YouTube could also improve the Content ID system through providing a better interface, more transparency, better compensation for artists, etc. Of course again that would mean lower profits for Google/YouTube so such straightforward fixes are unlikely. Meanwhile, YouTube makes great hay out of its concerns for poor, maligned users who may have received an erroneous DMCA notice. The company is willing to spend money to defend a few select uploaders but won’t spend resources to fix its broken Content ID system?
Operating only a marginal (not great) Content ID system is in YouTube’s best interests
Of course the powers that be at YouTube probably prefer to keep Content ID just the way it is–creaking along, occupying a neutral zone positioned between accolades and scorn. It’s a safe position, one that gives YouTube officials cover when they use disingenuous excuses about their anti-piracy practices to critics, while avoiding any real (legal or financial) consequences.
Content ID does the job just well enough….but that doesn’t mean it does a good job. It could serve as a true model for technological safeguards against piracy, but as now, it’s merely a slight bump in the road for those determined to steal and monetize the works of others. Meanwhile, YouTube continues to pocket advertising cash, make its stockholders happy while leaving filmmakers and musicians on the outside, looking in.