The pace of change has shattered old PC buying models.
Mark Schlack, Editor in Chief
Moore's Law has finally pushed the pace of PC hardware development through the sound barrier. A product manager for one of the Big Three PC companies recently pointed to one of his latest systems and told me, "In 12 months, everything in this machine will be different." And he meant everything: processor, memory, video, bus speed, network card, disk drive interface, and high-volume storage modes.
"Where've you been?" I hear some of you saying. "It's always been this way." To a degree, yes, but I'd submit that the increasing quantity of change has finally turned into a different quality of change. While everything in the aforementioned computer will be different in 12 months, the changes will happen continually, not just annually. What ever happened to the idea of a computer design you could settle on for a year?
Intel, which is driving much of this change, concedes that many new technology developments are pushed forward by intra-industry competition, not customer demand. But it's not all technology for technology's sake -- if you're a smart shopper, you can increasingly find just what you need among all the variants of PC technology. Buy that financial analyst a $3500 dual-processor system if she needs it, but don't shrink from spending a mere $999 on a system for the receptionist to get his e-mail. Got a dog of a Web server? Faster multiprocessor hardware's cheap compared to losing customers.
Nonetheless, Intel's attitude has a certain run-it-up-the-flagpole-and-wait-until-they-salute quality. Witness universal serial bus (USB) and MMX, two technologies that have waited long for third-party support. When one-third of your three-year life cycle is spent waiting for the very thing you bought a system for, can the words "inve
stment protection" have any meaning? After all, if you'd waited those 12 months, last year's $2400 200-MHz Pentium MMX would have cost you $1200. So before you buy a new technology, ask yourself if it's really compelling.
Compelling, to me, means at least the following:
- A technology must have an impact on the specific needs of users. Now that we're hitting 300-MHz clock speeds, for example, raw power alone is unlikely to be compelling for mainstream business users. Unless office suites go on another binge, users have clock cycles to burn. But if you're planning to implement voice recognition within a year, then horsepower is critical. In the end, is there a business case for the latest PC technology?
- A technology must be compelling in that it will change, within a reasonable time frame, the way you do computing. Power management was such a change for notebooks; the various management features built in to PC98 motherboards will be such a change for some shops. While you don't want to get
stuck buying the last vestiges of a technology (tried to buy an ISA video card lately?), there's no longer a coherent case to be made for habitually riding the front part of the wave. I'm no surfer, but I hear they give points for choosing the right wave.
So planning for the future has definitely become more difficult. In the coming months, we will be giving you some new tools to analyze the likely shelf life of systems for future applications from a performance perspective. But you'll have to add to that your own calculus of business need, tolerance of (or hunger for) change, and ability to support multiple technologies.
In the long run, the industry needs to find a saner way to introduce technology. We recently did a live survey at Comdex in which 74 percent of respondents felt that technology is moving too fast. But you're not powerless in this situation. When the industry tries to force a new technology before you're ready, take the bull by the horns and let system vendors know which en
hancements you can live without. In today's build-to-order world, the message will be received quickly.
Mark Schlack, Editor in Chief,
mark.schlack@byte.com
Coping with Change
