Watermarks: A better DRM than DRM itself?

If you're a digital content expert or you've encountered my series of blog posts on Digital Restrictions Management (DRM), then you should know by now that the "R" in DRM is officially for the word "rights" and not the "Restrictions" that I have been using as a permanent substitute.  But, in its current incarnations from various vendors, DRM technology has turned out to be more about stripping you of your rights than it is about managing them in a way that we'd consider management.  DRM is more restricting than it is right-giving or right-managing.  Even artists (eg: musicians) -- the very people whose rights DRM is also rumored to manage -- are outraged by the chilling, restricting effect of DRM.  But maybe, just maybe, there's a glimmer of hope in yesterday's announcement by TiVo.  But don't be surprised if the authorities come knocking on your door. If TiVo's "watermark" approach to the same problem that DRM was meant to solve takes any root, I'm wondering whether most if not all of the problems created by DRM could be eliminated overnight.  For the record, TiVo's public relations department has so far not responded to my request for more information about its announcement.  So, even though there's some guesswork in my analysis, I think you'll find the conclusion to be quite sound.

Some history is in order.  TiVoToGo is a technology from TiVo that basically allows the company's customers to take the recordings they make with their TiVo digital video recorders (DVRs) on the run.  For example, let's say you "TiVo'd" an episode of Desperate Housewives on Sunday night and want to watch it on the train ride into town on Monday morning.  TiVoToGo makes this possible.  In June of this year, Microsoft and TiVo jointly announced the TiVoToGo technology.  The connection between the two companies was quite clear.  TiVoToGo made it possible to take your recordings on the run with a portable playback device but, outside of DVD players and PCs (also supported by TiVoToGo), it only works if you're on the run with a Microsoft playback device.  Your Microsoft playback device.  Under the hood, it was probably a combination of Microsoft's Windows media format and DRM technologies that restricted playback to such devices (I say probably because I can't confirm this with TiVo).  At the time, there was no video iPod.  But there were Microsoft-based PDAs, smartphones, and mobile content players (MCPs) that were enabled to playback video content "wrapped" in Microsoft's formats and DRM. 

Despite what naysayers have to say about looking at small displays, taking recorded video on the run is a great idea.  Given the choice between nothing and a small display, many people will  take the small display. So, TiVoToGo as a concept makes perfect sense.  But, along with that sort of portability of content (from DVR to MCP) comes the risk that the content will end up being pirated into unintended distribution channels. By restricting  the "ToGo" part of TiVoToGo to specific playback devices, TiVo was doing its best to prevent such leakage and business model disruption.  But, in doing that, there were other unintended consequences.  With more MCPs coming out, many of which are not compatible with Microsoft's technology (eg: Apple's video iPod), TiVo was denying itself access to the entire market. 

To rectify the problem, TiVo has one of two choices.  One of those is to turn TiVoToGo into a multiplexing technology that's smart enough to "portabilize" content using the format and DRM scheme that matches whatever MCP the end-user has.  For example, if it's a Microsoft PlayBack device, use Microsoft's video formats and DRM.  If it's a video iPod, use Apple's video formats and DRM (the two are completely incompatible with each other).  If it's a Sony player, use Sony's formats and DRM.  You get the picture.  It's a lot of effort -- fraught with the sort of technical complexity that makes rootkits look simple -- that borders on  the absurd when it comes to what TiVo should have to do to make its customers happy.  Not to mention the technology licensing issue. 

Not only are royalties (costs that a company facing stiff competition the way TiVo is shouldn't have to bear) involved when building a device that is capable of encoding content with another company's file formats and DRM technology, not every company is licensing its technology as liberally as Microsoft is.  With the exception of some Motorola phones, Apple -- a company that, given the way it has its sights set on being the digital media hub,  is a likely candidate to enter the DVR business at some point in the future -- has kept its technology to itself.  Given the popularity of iPods and iTunes Music Store purchased content, licensing the technology to TiVo could undermine Apple's own success in the digital media business. Sonos, a maker of a wireless audio solutions that compete with Apple's AirTunes, is all too familiar with the exclusionary tactic.  As I've written before, although Apple does not have a monopoly, I believe this behavior to be monopolistic.  

The other choice TiVo had was to try for something much simpler and cheaper for the company to manage. Go with a format that virtually all digital video playback technologies understand, get rid of the DRM, and find another way to send a clear message to its customers that the illegal redistribution of TiVo'd content will not be tolerated.  Exit DRM and enter watermarks.  To make its recordings compatible with MCPs of all types (including Apple's video iPods and Sony's PSPs, both of which are mentioned in TiVo's press release), we have to assume (again, no call back from TiVo) that instead of converting to Apple and Sony's proprietary formats, TiVo will be converting to the more widely supported MPEG-4 video encoding scheme.  Coverage of the news by Engadget suggests this is the case.  Just about anything capable of playing back digital video can playback an MPEG-4-based video.  But MPEG-4 has no provisions for DRM, and as such, practically greases the wheels of content piracy. This is where watermarks come in.

Here's what TiVo's press release has to say on the issue:

Subscribers will need to purchase certain low-cost software to facilitate the transfer of content from the PC to these portable devices. To discourage abuse or unlawful use of this feature, TiVo intends to employ "watermark" technologies on programs transferred to a portable device using the TiVo ToGo feature that would enable tracking of the account from which a transferred program originated.

And here's what Engadget said in its coverage about the watermark feature:

...recorded shows will be digitally watermarked allowing content to be tracked back to your living room if you get torrent happy with ‘em, dig? Fair enough, as this still allows us to view the content on as many portable devices as you like...

In other words, content that you download and encode will be uniquely watermarked so it can be traced back to you. So, if you want to load something you recorded up to the Net for illegal distribution, go right ahead.  But don't be surprised if the authorities come knocking on your door a few days later. At first glance, a lot of people will say this sort of system sucks.  Yeah, it does.  It sucks that content providers are forced to go through such hoops and content consumers like us are forced to put up with those hoops.  But, on the other hand, compared to how restricting DRM is, this is infinitely better because it focuses on rights instead of restrictions.  It gives us the rights we should have to view our content anytime, anywhere, on any device we want.  Not where TiVo says we can view it. Not where Apple, Microsoft, or Sony says we can view it.  It even enables us to illegally upload that content at the Net.  But, instead of restricting us from doing that, it basically sends the message that if we do so, we do so at the risk of our own peril.  The burden is shifted away from ridiculous friction-inducing technologies and back to the content providers who must seek out and destroy illegal distribution channels.  Only this time, there's a smoking gun (the watermark); one that you really don't want to leave behind.

Of course, one of the reasons that watermarks can be so effective is that -- with all those pixels and colors that go into every frame of digital video -- watermarks are easy to bury into video in difficult to hack ways that don't substantially interfere with the viewing experience.   This raises the question of whether or not watermarks can be equally effective for audio.   For example, if you add bits to an audio file, will they turn up in the listening experience? If the answer is yes, might audio watermarks involve some compromises that make them easy for hackers to strip out? Via e-mail, I interviewed an expert on the matter -- Tony Faulkner, partner at the UK-based Green Room Productions.  What makes Faulkner an expert on the topic is that Green Room Productions produces digital recordings of classical music and recordings of classical music are probably the ones that can least tolerate unwanted noise: the sort of noise that a watermark could potentially introduce to such a recording.  Faulkner is a member of the Audio Engineering Society's Technical Council and recently chaired a meeting on audio watermarking in Los Angeles that  involved some of recording industry's biggest players.  With his permission, I've posted a complete reproduction of his very thoughtful e-mails to me here, but the short of it is that audio watermarks can pose challenges to the most sensitive of listeners -- people who the audio industry refers to as "golden ears" -- but they're challenges that Faulkner believes can be overcome.

After responding to my series of DRM and issuing a warning that Sony may have something even more insidious in store for us than rootkits -- something called AACSLA --  Faulkner went onto discuss the potential for audio watermarking and why one previous attempt failed:

Watermarking as a concept is a good concept and used successfully in graphics, so long as it does not corrupt the material it is supposed to be protecting.  Superimposing a buzz of pulses slap bang into the middle of the analogue audible band of music is pushing your luck.  [If] it can be very, very low level and masked by system noises and programme, then you get away with it. It can be digital and put into the data-stream without sounding terrible.  However if you need the watermark to be robust enough to survive copying to/from analogue cassette via a telephone line in analogue, then the robustness requirement means it is more likely to be audible.......Watermarking digital audio should also be quite possible without degrading the quality irreparably.  The previous effort which the audio community got upset about was a different matter: it interfered specifically with the main audio band, which is daft for a high quality medium.  We were more angry about the quality degradation before any consideration of deprivation of consumers' rights.....My main concern now is not with DRM which degrades audio quality, it is with DRM constructed by incompetents and intended in bad faith to deprive legitimate money-paying customers of their rights.

Of course, bear in mind that a lot of the music that people listen to is so noisy that burying a watermark in those recordings would be almost as simple as burying a watermark in video.  Especially when you consider that most music is being played back from a compressed format anyway (in other words, the listener has already "accepted" certain compromises to the quality of the music).  To me, watermarking, as it apparently surfaces in TiVo recordings and as described to me by Faulkner, sounds like a far more promising approach to dealing with the piracy problem than is DRM.  Even if I don't all the details on the TiVo announcement right, the promise of watermarking still stands.  It maintains the rights of all those involved (rights holders as well as listeners) while creating a "system" that clearly deters illegal copying and distribution.  I'm sure there'd be some low level, below the radar pirating... but that's the sort of pirating that's always been a part of the recording business.  The costs of that have long been built into Hollywood's pro forma.   It eliminates the introduction of friction to a system that technologists should be looking to strip of friction and could lead to the added benefit of letting users decide for themselves whether they want to sacrifice storage space for quality (for example, going with a space consuming lossless audio format like FLAC instead of a very lossy but space saving one like MP3).

So, what are there pitfalls? Right now, I can't think of any.  Can you?