Watermarking & DVD Conspiracy Theories

I've recently been rereading Mark Lane's and Donald Freed's 1970s screenplay cum novel, Executive Action, which develops the theory that John F. Kennedy was assassinated by a conspiracy between organized crime, expatriate Cuban Batistists, and Eisenhower's "military-industrial complex." Long predating Oliver Stone's JFK, the book is fascinating, convincing stuff, from authors who had done considerable research into what really happened in November 1963. But, like all conspiracy theories, it falls down on the hard rock of reality: the more people and organizations are involved in a conspiracy, the less likelihood there is of anything happening at all, let alone going according to plan.

Thus it has been with the much-heralded launch of DVD-Audio. With so many entities involved in the development of this high-resolution multichannel music medium, progress has been painfully slow, particularly when compared with the speed with which Sony and Philips have brought to market the competing Super Audio CD. The preliminary specification for DVD-A was released for discussion as far back as the late spring of 1997; the adoption of Meridian Lossless Packing was announced in June 1998, and the DVD Forum's DVD-Audio Working Group 4 finalized the DVD-Audio 1.0 specification in October 1998.

But, as Meridian's Bob Stuart told me in August '97, although there was agreement among all the DVD Forum members on the key components necessary to ensure DVD-A's success in the marketplace—video, three-dimensional sound, and the highest possible fidelity—these were overshadowed by the recording industry's need for reliable copy protection. It is that need that torpedoed the format's 1999 launch.

As you can read in this issue's "Industry Update" (p.15), Matsushita (Technics, Panasonic), Nippon Columbia (Denon), and JVC were alarmed by the cracking last November by Norwegian hackers of the DVD-Video Content Scrambling System (CSS) encryption scheme. They decided to postpone bringing their players to market until the "4C Entity"—a consortium of Matsushita, Toshiba, IBM, and Intel—develops a new copy-protection scheme for DVDs (footnote 1). It was reported that, as a result, Nippon Columbia had to trash 10,000 discs that had been produced in time for the DVD-Audio launch.

The DVD-Video encryption involves each manufacturer identifying its players with a unique 40-bit string, or "key." Each DVD-V disc includes a short in-the-clear "disc key," which is also encrypted with all the allowed player keys. When you insert a DVD in a player, a considerable amount of handshaking takes place, the player using its key to descramble the appropriate encrypted disc key before playback is allowed. (Perhaps this is the reason DVD players are so slow to load.) The hackers got hold of a unencrypted player key and used it to reverse engineer the system. Their program, widely available on the Internet, allows anyone with enough computer storage space to make decrypted copies of the files that make up a DVD movie.

So now that CSS has been cracked, the DVD Forum had to choose between disabling all keys, thus turning the existing population of DVD-V players into worthless junk, or developing a stronger algorithm than CSS.

But why should the cracking of the DVD-Video encryption have had such a catastrophic effect on the launch of DVD-Audio? From what I've read, the CSS algorithm was never intended to be a strong deterrent to pirates. (Heck, a friend once saw a pirated video that had been made by someone pointing a camcorder at a cinema screen!) Yes, the DVD-V protection may have been hacked, but only a few computer geeks will have the time, patience, and skill to take advantage of this.

Common sense says that the primary protection against piracy is low pricing: At current DVD-V prices, Joe Sixpack has no motivation to do anything other than buy legal discs. And, as Barry Willis points out in this issue, common sense also dictates that satisfying the need of the movie and record industries to protect their copyrighted material is a chimerical dream—no matter how robust a DVD's encryption, the unprotected datastream must eventually be exposed in order for the movie to be viewed and the music to be heard.

But when so many entities are involved, it becomes hard to get agreement on anything, even if it's just plain common sense. And the record industry may well have more to fear than the movie industry. The sheer amount of data represented by an MPEG-2-encoded movie makes a DVD intrinsically hard to copy, but a single cut off a DVD-A, even if multichannel and 24/96, is an easily manageable amount of data for anyone with a modest computer to copy. According to Matsushita spokesman Yoshihiro Kitadeya, as reported in a Dow Jones Newswire story on December 2, "since music requires much less data than video, it would be easier to copy individual songs from a DVD-A disc...and distribute them cheaply over the Internet."

That's the real danger behind the music industry's paranoia—that the DVD-V hacking will make it easier for people to distribute their songs on the 'Net. Being paranoid doesn't mean you don't have people out to get you, and it explains the music providers' insistence on encryption, copy protection, and "watermarking"—the embedding of coded identifying data in the music that is sufficiently robust to survive successive D/A and A/D conversions, and even the transmission of analog versions of the music over AM radio.

One of the more depressing experiences I had recently was attending workshops and paper presentations at last September's Audio Engineering Society Convention on the subject of watermarking. The record industry's need for the level of protection to be "robust" means that the watermark must be high in level. The only thing preventing its immediately being heard is that it is encoded to mimic the recording's noise floor, frequency-shaped so as to be masked as much as possible by the music's spectrum. Yeah, right. And now, according to another "Update" story (p.17), TTR Technologies and Macrovision are developing a copy-protection system to prevent "casual copying" of music CDs. In my opinion, the balance between music-lovers and copyright holders is leaning far too much in the direction of the latter.

When it comes to higher-resolution music, my advice is therefore to ignore the ill-fated, protection-hobbled DVD-A. Buy instead the DVD-Video–based DADs from Classic Records and Chesky, or invest in one of Sony's Super Audio CD players and the SACDs to play on them—unless, of course, the music industry belatedly decides that even SACD offers inadequate protection against copying.

Footnote 1: Late-breaking news at the time of writing this column was that Pioneer intends to go ahead with its late-December launch in Japan of two DVD-Audio players: the high-end, $5000 DV-AX10, which will also play SACDs; and the less-expensive DV-S10A DVD-V/A player. Pioneer points out that while no DVD-A software exists, these players will play DVD-Vs, CDs, and SACDs (DV-AX10), and that purchasers will be able to get their machines upgraded when a new encryption scheme is announced. In the meantime, Pioneer might also release copy-protection-free DVD-A discs; the company has already released a DVD recorder!