I've been down on MPEG-4 for a while now, despite my respect for many of the folks associated with the standard. Given the adoption of Microsoft's Windows Media Video 9 by the DVD Forum, there's increasing reason to believe, to paraphrase an old Southern expression, that the MPEG-4 dog just won't hunt. It hasn't yet and probably never will, at least in any serious commercial way.
What spurred this belief into this column was a quote I read in High St@kes, No Prisoners, a book by Charles H. Ferguson, who founded Vermeer Technologies, the FrontPage developer purchased by Microsoft for $130 million back in 1996. The book is fascinating on a number of levels: the hunt for financing, brushes with Netscape and other dot.com high fliers, and, of course, the story of the Microsoft acquisition.
In the book, the professorial Ferguson delineates how software architecture is key to producing stable software and how standards impact the marketplace. Are standards better set by so-called monopolists (e.g. Microsoft) or standards bodies? On the last issue, he writes:
"The battle to create and own a proprietary industry standard generates rapid improvements in price and performance, at least until somebody emerges totally dominant. If fact, even after a monopolist emerges, there is still considerable pressure to innovate, because unless you can induce your installed base to upgrade frequently, you have a hard time continuing to grow. There is a risk that a monopolist can misbehave, but slow moving, least-common-denominator, nonproprietary, committee-developed standards are often much worse than any monopoly."
That last sentence kind of describes MPEG-4 to a T, doesn't it? Of course, your first thought might be, "Well, what about MPEG-1 and 2, two standards that have succeeded beyond any reasonable expectations?" Great question, so I'll address them first.
Simply stated, MPEG-1 and 2 were developed almost in a vacuum, with no real competition in the computer marketplace. In fact, they were developed almost exclusively for non-PC devices—MPEG-1 for VideoCD and CD-i (DVD-like players that connected directly to television sets for playback), and MPEG-2 for broadcast and DVD.
At the time, with computers going through that rough, 80486, Windows 3.1 stage, PC-based compression technologies danced the bitter compromise between quality and display rate. Whereas MPEG-1 could count on (and required) dedicated hardware for playback, Indeo and Cinepak had to play on the computers of the day. Definitely no competition for MPEG-1 since their focus was entirely different.
When MPEG-1 was first introduced into the personal computer space, it was via upgrade kits that had a failure rate exceeded only by the Boston Red Sox's post-Bambino attempts to win a World Series. Three or four years later, the initial attempts to integrate MPEG-2 into computers via hardware playback accelerators met with much the same fate.
In fact, both MPEG-1 and MPEG-2 owe their entire success in the PC space to Moore's Law, which boosted computer performance sufficiently to play both formats without dedicated hardware. As we'll see in a moment, what Moore's Law can give, it can also take away, a prime factor in MPEG-4's imminent demise.
Contrast this to the current streaming media environment. Before MPEG-4 was introduced, three streaming technologies enjoyed widespread use: RealVideo, Microsoft's Windows Media Video, and Sorenson Video. MPEG-4 offers lower quality than any of them.
In addition to offering lower quality, MPEG-4 also sports the obligation to pay royalties, not only on encoders and decoders, but also on content. Given that decoders for the other three have always been free, it's doubtful that this costs more, looks worse "value proposition" will win many takers in the streaming media space.
This leaves appliances like DVD players, cell phones, and PDAs as the last hope for MPEG-4, but Moore's Law will prove its undoing. Specifically, in the past, these devices were driven by single-purpose chips that could only be prudently designed and manufactured for a single, well-supported standard.
Today, many of these devices are built around general purpose digital signal processors (DSPs) that can support multiple compression technologies. Since these chips aren't locked into a specific technology, this lessens the importance of standards in the technology decision.
Finally, it's hard to minimize the importance of the DVD Forum's provisional approval for Microsoft's VC-9 technology, essentially Windows Media Video 9, along with two other technologies, H.264 and MPEG-2, as mandatory on next-generation playback devices.
This announcement has three significant implications. First, it forces the committee-driven standards to compete primarily on performance, something neither has had to do yet. Second, it opens the door for proprietary standards to compete against committee standards in other markets, whether digital television or Dick Tracy-like video watches. The combined impact of these realities will either force committee members to work harder and faster or spell the death of committee-based video codec standards.
Finally, it presents Microsoft with an economic justification for continuing to invest in Windows Media Video technology. To date, Microsoft's obvious investment in video technology and infrastructure has been made for strategic reasons, not for profit. The promise of DVD-related license fees should spur development even more.