There are plenty of features worth tracking, but taken as a whole they are not likely to boost the PC industry far beyond the 8 percent compound growth and 7 percent decline in average prices expected over the next four years.

As usual, Microsoft has no shortage of small, innovative design teams run by aggressive thirtysomethings, all trying to pack cool features into Longhorn. Hard to tell which strands of this spaghetti will stick to the wall.

Perhaps the most intriguing bit of work is a new architecture for copy protection that Microsoft hopes will convince broadcast and content companies to let digital TV and high-definition DVD content flow on the PC. Another team promises to upgrade audio and video performance on Longhorn PCs. It is developing a scheduler, a heap manager and fresh techniques for managing threads and storage. A related team is developing a whole new user-mode audio stack for Longhorn, promising professional-quality audio that will compete with consumer systems. It works, in part, by tapping into 500-microsecond timer resolutions in the kernel and by loading code and data into nonpageable memory. The promise is for a signal-to-noise ratio of 30 to 144 dB using XP-class CPUs and memory.

Yet another group is working on core performance, using a caching technique it calls superfetch and a fresh approach to how Windows uses flash and moves files on and off the disk. It's not clear whether this same team is working on the new, fast access techniques that Microsoft claims will let Longhorn systems boot from disk in just 15 seconds and come back from memory sleep states in 2 seconds.

There will also be a new Internet Protocol stack, with deeper support for IPv6, Microsoft said. And Longhorn will automate the way systems find and set up secure wireless connections and automatically move from one net to another. Finally, a 3-D user interface will bring a bit of sizzle to Longhorn, leveraging scalable vector graphics with pixel shading.

But there are also weaknesses. Microsoft scaled back an ambitious program called Next Generation Secure Computing Base (NGSCB) for putting a hardware-based root of trust in Longhorn PCs. Instead, it is commercializing a much simpler feature, called Secure Startup, to encrypt a hard drive automatically so that the drive is useless if someone breaks into a computer or finds a lost system.

NGSCB was born back in the days when Microsoft was getting hammered by both Hollywood and the Department of Homeland Security about the lack of security on the PC. With government pressure easing and a simpler, software-based idea emerging for the film industry, the company decided to phase in the NGSCB vision over a longer period. But even some of Microsoft's closest NGSCB partners say they don't know what the security plans are beyond Longhorn.

Similarly, Microsoft backtracked on plans for a new database-style file system. Instead, it will put some interesting capabilities into Longhorn for searching data across applications, with richer indexing and metatagging. The prototypes look good, but Google and Yahoo desktop products may not be that far behind.

Microsoft has started an "ultramobile" project called Haiku. The target is a full PC weighing less than 2 pounds and sized slightly smaller than a steno notebook, with a 3G cellular radio, camera and touchscreen built in. It would play music and movies, run more than 6 hours on a battery charge and sell for less than $1,000 when it hits in 2007 or so.

This could be just another misfired PDA, but chairman Bill Gates is big on it. Bill Mitchell, the Microsoft mobile vice president who ran the Windows CE and PocketPC programs, is driving the project.

Intel Corp. has tipped plans for something along the same lines. Its On-the-Go PC weighs 2.5 pounds, comes with a DVD player and has an innovative 8-inch screen that acts as a speaker. It will sell for less than $1,000 this year, based on Intel's Centrino and a shrink of its latest chip set. This space bears watching.

Threats

Longhorn arrives at an awkward moment in the evolution of wireless networking and packaged media, two of the hottest vectors in computing. Standards for ultrawideband, 802.11n and WiMax short-, local- and wide-area nets are being hammered out even as Longhorn approaches first beta release this summer. At the same time, a consolidation of high-definition DVD formats is in the works.

Thus, solid support for the new wireless and DVD features will not be natively baked into the next OS, at least not in its first incarnation. Microsoft fully appreciates the awkwardness of the situation and is responding as best it can. Nevertheless, the bad timing will create some time-to-market and interoperability bumps in the road for the first Windows products to build in those technologies.

Specifically, Microsoft is rewriting its Wi-Fi stack and creating a wireless "extensibility framework" so that system and chip makers can add in their own code supporting .11n, WiMax and UWB products. Microsoft has also joined the WiMedia alliance to get closer to the UWB standards process.

Some observers here predicted a replay the fiasco that occurred when Windows XP came out without support for Bluetooth. But given the strong focus Microsoft and Intel are putting on UWB, don't expect that.