Media Wonk

Link This | Email this | Comments (0)

More fingerprint filing - October 16, 2007

Hard to tell yet whether YouTube's proposed copyright filtering technology will get beyond the beta stage.

As YouTube itself acknowledges, the system it's designed requires considerable cooperation from content owners to be effective. Essentially, a studio has to provide YouTube and its parent Google with a complete copy of any work it wants filtered, as well as a detailed outline of how it wants YouTube to respond to discovering a piece of protected content on the site: block it altogether, let it go but attach an ad to it, permit a particular use so long as it's not too much, take it down after a certain period, etc.

The system seems designed, in other words, intentionally to keep as much of the labor and cost burden of filtering as possible--as well as the legal liability--on the content owners. Whether they'll really bite on that remains to be seen.

One thing that is clear though is that should the system go forward, Google will eventually compile a huge and immensely valuable database of video content and fingerprints, which will inevitably raise the question of who will control it, and what will they be able to do with the database.

"It has the potential to be an industry-changing event," Teletrax president Andy Nobbs told Media Wonk. "For the first time, you'll have a database that can identify a piece of video anywhere on the Internet, something you really can't do at all right now. That means you'll be able to market to people based on what they're watching, develop profiles based on their viewing habits. It would be hugely valuable."

Teletrax provides digital watermarking technology to content owners for tracking video usage on television. But its system, like most tracking technology, looks for bits of code inserted into the video to identify specific pieces of content. YouTube, in contrast, is proposing to generate fingerprints from randomly sampled pieces of video and then compare those against a vast database of complete works looking for a match.

"If it was anybody besides Google doing it I'd say it couldn't be done," Nobbs said. "I assume they can throw a lot of servers at the problem."

The tricky question will be legal ownership of the information in the database. Clearly, the works themselves belong to the original copyright owner. But what about the fingerprints Google plans to generate? They refer to works owned by the studios and TV networks, and might in some sense be considered derivative of those works. But they're not reproductions of the works, or replacements for them. They're data about the works.

The question has come up before, in the 2001 case of Gracenote v. Roxio. In that one, Gracenote was sitting on an open-source compilation of digital music fingerprints, which, for awhile it allowed companies like Roxio (EasyCD Creator) to incorporate into their applications free of charge. In 2000, however, Gracenote sought and received a patent on the database and its use in certain systems and went after Roxio and others for license fees.

The case was settled before trial, and the settlement agreement included a stipulation by Roxio that the Gracenote database "includes protectable copyrighted subject matter."

I can't imagine the studios agreeing to that point, especially if they're providing the raw material that goes into the database. But Google may think otherwise.

Should be fun.
[Content Protection & Management]  [Digital Copyright]  [DRM]  [Legal]   LEAVE A COMMENT