Under US copyright law (17 U.S.C. § 512, or "512" for short), a content owner can send a message to the operator of a network service like YouTube if it finds an item of its content on the network, and the network operator is required to take the content down; this is called notice and takedown.  YouTube's position is that it complies with the law by responding to takedown notices and by offering tools to content owners that automate this process -- presumably such things as web forms or email message templates for them to fill out. 

On the other hand, the position of content owners is that notice-and-takedown is reactive -- it only takes place after the work is up on the network and the damage is presumably done -- and that it places too much of a burden on them to find works and issue the notifications.  In content owners' ideal world, their works would never make it onto these networks in the first place (without their authorization), and they would not have to lift a finger to make that happen. 

Enter fingerprint filtering, the technology that content owners view as enabling that ideal world.  The technology identifies content when a user is trying to upload it, and if it's a work in the technology vendor's database, it blocks the file from being uploaded (or takes some other action according to the content owner's wishes).  A handful of vendors have implemented this technique for music, including Audible Magic, Philips, and Gracenote. Analogous techniques for video are being developed.

The major music companies generally endorse fingerprint filtering and have licensed their content to services that use it, such as iMesh.  MySpace uses it to block music uploads.  At the same time, the question of how well the technology works in large-scale production applications is still somewhat open.

The key difference between notice-and-takedown automation and fingerprint filtering, of course, is that the former is reactive while the latter is proactive.  But the other key difference is in the burden that each technology places on content owners and network operators.  Viacom has complained about the cost and time it takes to search services like YouTube for its own content in order to give notice under 512, whether through automation or not.  A network service provider can cut its costs of responding to 512 notices by offering automation -- though that begs questions of whom to trust if someone gives notice on content that they don't own, and what happens if that occurs and a work is inappropriately taken down (as the EFF has suggested in Viacom's case).

With fingerprint filtering, most of the burden shifts to the network service provider and the fingerprinting vendor it uses.  The content owner needs to feed its content to the fingerprinting vendor for fingerprint registration, but it only needs to do that once for each content item -- a task that can become part of the process of product release. 

So, as usual with DRM-related technologies, the issue becomes one of who is going to pay: content owners or downstream links in the value chain.

We could suggest a couple of ways to partially bridge the gap between notice-and-takedown automation and fingerprint filtering.  One is to standardize a way to automate notice-and-takedown, so that content owners can send the same types of messages to many network service providers instead of having to deal with each one differently.  Content owners' trade associations, such as the RIAA, could define a standard protocol for conveying 512 notices; this would be perfectly acceptable under antitrust law.  This would make it more efficient for each content owner to send 512 notices to multiple service providers (e.g., via RSS feeds) and thereby reduce their burden a bit, making notice-and-takedown slightly more acceptable to content owners.

Another idea might be to standardize on a single fingerprinting technology and make fingerprinting a part of copyright registration.  Although US law does not require copyrights to be registered in general, it does require registration as a prelude to infringement litigation.  A neutral entity such as the US Copyright Office could maintain a database of fingerprints, and content owners could submit their content for fingerprinting when they register for copyrights. 

There are several issues with this scheme, though.  First, what would motivate network service providers to use the standard fingerprint database to block unauthorized uploads?  From a technology standpoint, the answer would be the confidence that the media industry backs a standard solution and the lack of need to go around to multiple content companies to get their blessing on a nonstandard one.  Presumably the technology would be designed with robust and convenient interfaces, so that network service providers could integrate with it as easily as possible.  From a legal standpoint, content owners are hoping to argue that not implementing filtering could be tantamount to inducement of copyright infringement, which the Supreme Court's 2005 Grokster ruling defined as criminal. 

Another major issue with fingerprint standardization is how it could be done in a way that satisfies antitrust concerns and ensures that the technology could be developed over time to be as robust, accurate, secure, and scalable as possible, given the inevitability of hacks and workarounds.

The final issue is who will pay for it.  The simple answer to that one seems to be that content owners pay a surcharge for fingerprinting when they register their copyrights.  But this is not so obvious, given that even if network operators were to share the cost, it would save them money compared with each implementing their own fingerprinting schemes. 

Deliberation between notice-and-takedown automation and filtering is certainly not the only bone of contention between the media industry and content service providers these days, but it's an important one.  How it gets solved -- and to what degree the legal system mandates those solutions -- is an important barometer of progress (or lack thereof) in the online content arena.

Get DRM Watch Newsletter

Click here to subscribe to DRM Watch