The PC: Imperfect--and Indispensable

PCs have flaws. But the machines work well enough for most--and they're cheap

Twenty years ago, a rather reluctant IBM (IBM ) unleashed the homely Personal Computer 5150 and created the business PC. Since then, speed has increased by many orders of magnitude, the basic unit of storage has gone from the 160-kilobyte floppy disk to hard drives that hold tens of billions of bytes, and the price per unit of processing power has fallen perhaps a millionfold. Yet the fundamental architecture of the PC has changed amazingly little. In fact, most programs and hardware accessories that worked on the original IBM PC still work today. Plug an early 1980s Epson MX-80 dot-matrix printer into a Pentium 4, and it will be ready to chug away. What accounts for this longevity? And what does it portend for the future of personal technology?

Probably the most important lesson is that technological elegance isn't terribly important in the marketplace; being good enough will do. IBM's requirement that the PC be built from cheap, off-the-shelf parts forced engineering compromises that persist today. Over the years, techno-snobs have derided PC hardware and the software that runs on it, with Sun Microsystems (SUNW ) CEO Scott McNealy famously dismissing Microsoft Office as "a hairball." None of this has mattered. "It wasn't rocket science, and it wasn't extraordinarily inventive," says IBM Fellow Mark Dean, who designed the color-graphics adapter for the original PC. "Most of the time it's not the level of technology that makes a difference. The important thing is that you are solving a problem at the right time and at the right price."

FIERCE RIVALRY. The PC became a vital business tool after Lotus Research introduced the 1-2-3 spreadsheet in 1983. At about the same time, IBM, which never tried very hard to retain exclusive control of the basic design, lost control of the technology to Microsoft (MSFT ) and Intel (INTC ). The consequence has been fierce competition among PC makers--and a steady slide in prices.

The first real challenge to the PC's dominance was the introduction of the Apple Macintosh in 1984. The Mac was and is technically more elegant than its Intel-Microsoft rivals. But it was a closed system, compared to the PC's competition-inspiring open architecture, and Apple (AAPL ) claimed a price premium. By the time Microsoft narrowed the usability gap with the introduction of Windows 3.1 in 1991, Apple Computer's chances of making significant inroads were over. The PC wasn't the best, but it was good enough.

In the early 1990s, the PC reigned unchallenged. But by the second half of the decade, the growth of the Internet created new competition in the form of "network computers" and, more recently, "Web appliances." These were simple terminals that depended on the processing power and intelligence of network servers. Despite the backing of such heavyweights as Sun, Oracle, and IBM, network appliances failed miserably. The reason is no mystery: They delivered a small fraction of the performance and flexibility of even a low-end PC at a large fraction of the cost. Another challenge to the PC has come from handheld devices. But too many tasks depend on a good keyboard and a big display for these pocketable tools to be more than adjuncts.

I have been as vocal as anyone in criticizing the PC's many flaws. The computers are still too hard to use. Software quality leaves much to be desired, and consumers are expected to put up with a level of failure and frustration that they would not accept from any other product. Yet I have to admit that the reliability of the hardware and the usability of the software have improved dramatically over two decades.

In one sense the PC's glory days are over. Many markets are largely saturated, and even a return to rapid economic expansion is unlikely to produce the sort of PC growth rates we saw in the late 1990s. In the corporate world, researchers are working on ways to turn computing power into a utility in which clusters of servers are used to sell computing power on demand, sort of the network computer idea on steroids. And eventually someone will design a Web appliance that will turn consumers' heads.

Still, the value proposition of the PC--good enough and really cheap--will make it hard to displace. I wouldn't be surprised if 20 years from now, offices and schools were full of Pentium 15 PCs running Windows 2020.


    Before it's here, it's on the Bloomberg Terminal.