The PC at 20
Q: How do you crash Microsoft Windows?
A: Start Microsoft Windows.
The joke has been around for years. There are plenty of reasons why Windows PCs are so unstable. And one of the biggest is history--20 years of it, now. Today's Wintel (shorthand for Windows on an Intel processor) systems descend directly from the original IBM PC. Announced in August 1981, it was a product that IBM rushed to market with no inkling that it was setting a worldwide standard that would prevail for decades to come.
Of course, all that history has its good side. IBM's machine was extremely versatile, infinitely upgradable, and well documented--strengths that led to the PC's initial success and enduring ubiquity. But those same virtues led to a haphazard evolution that piled one problem on top of another.
With their beautiful graphics, multitasking applications, and networking talents, today's gigahertz-plus systems seem a far cry from the PCs of two decades ago. Still, at the heart of every 21st century Windows-based computer lies an IBM PC.
"What's amazing," says Dan Bricklin, whose Visicalc was the first PC spreadsheet, "is that you can take software for the original PC and run it on today's Windows [systems]."
Since 1981, PC technology has seen remarkable advances--and more than a few false starts and outright blunders. So let's look back and see how today's systems got the way they are. Return with us now to the dawn of PC history . . . .
If you were in the market for a personal computer back in 1980, you had plenty of choices. Many popular models of the day ran an operating system from Digital Research called CP/M. Commodore's PET and Tandy/Radio Shack's TRS-80 were also established players. And two guys named Steve had a big business selling the Apple II.
Those systems were aimed at hobbyists who liked to write programs in the BASIC programming language. The major supplier of BASIC was Microsoft, a little company in Bellevue, Washington, headed by a Harvard dropout named Bill Gates.
IBM, the biggest name in serious--that is to say, large and business-oriented--computers, took notice of the nascent personal computer industry in 1980 and assigned a veteran manager/engineer named Don Estridge to get an IBM PC to market. Under strict deadline pressure, IBM engineers in Boca Raton, Florida, made decisions that are still with us today.
David Bradley, who was a member of Estridge's engineering team, recalls, "The
system would hang, and the only way you could fix it was to turn it off. So I
built a warm reboot into the keyboard code. I invented
The tight deadline meant that the system had to be built with existing technology. Its central processing unit was Intel's 8088. The 8088 was a 16-bit chip--a zippier, more powerful CPU than the 8-bit CPUs used by most early microcomputers. But to keep costs down, the 8088 talked to other components via an 8-bit bus.
Why didn't IBM stick with the popular CP/M operating system? Legend has it that Digital Research president Gary Kildall skipped a meeting with IBM execs to go flying or hang gliding. A more likely story is that Digital Research refused to sign IBM's nondisclosure agreement.
One fact is undeniable: Digital Research had yet to deliver a version of CP/M for Intel's 16-bit CPUs. So Tim Paterson, an employee of a small hardware vendor called Seattle Computer, wrote a CP/M-like operating system for that company's computer, which used Intel's 8086 processor (basically an 8088 with a 16-bit bus).
"I didn't have time to do it right," he recalls, "so I did it quick and dirty." In fact, his creation, officially called 86-DOS, was nicknamed QDOS--Quick and Dirty Operating System.
When IBM told Bill Gates about its problems with Digital Research, Gates had a solution. Microsoft acquired a license for, and later bought, QDOS from Seattle Computer, then licensed it to IBM. It was renamed IBM Personal Computer DOS (or PC-DOS) when sold by IBM, and MS-DOS if sold by anybody else.
But when the PC hit the market, PC-DOS was one of
Nobody was prepared for the IBM PC's instant, explosive success. And that clamor was for a machine whose $1265 base model didn't include a monitor, a video card, a parallel or serial port, an operating system, or a floppy drive. According to Bradley, IBM hoped to sell 241,683 PCs over five years. Before those five years were up, the company was selling nearly that many units a month.
And certainly no one expected the standard to last for decades. The PC industry "wasn't seen as having a present, much less a future," remembers Mitch Kapor, creator of Lotus's 1-2-3 spreadsheet.
Why was the PC a hit? For starters, it was a well-designed, well-built machine from a name that businesses knew. Clever ads, with an actor imitating comedy legend Charlie Chaplin, also helped.
And the PC was quickly supported by a raft of third-party applications, such as word pro-cessors (MicroPro's WordStar, SSI's WordPerfect, and others) and Ashton-Tate's DBase database manager.
Peter Norton developed
the first version of Norton Utilities to restore a file that he'd accidentally
deleted on his own system. And Andrew Fluegelman invented shareware with
PC-Talk, a program that made it relatively easy for modem users to dial in to
services such as CompuServe and the Source. (Later, Fluegelman helped start a
magazine you may know:
But it was Kapor's 1-2-3 that cemented the IBM PC's reputation as a business machine when the program debuted in early 1983. By taking advantage of the personal computer's 16-bit CPU, 1-2-3 could offer revolutionary features such as on-screen menus.
IBM's Estridge--who was to die in a plane crash in 1985--and his team wanted other companies to supply a broad selection of peripherals, so they designed the PC as an open, well-documented system. They got the peripherals, but they also got something else: clones.
In 1982, a start-up called Compaq released a "portable," sewing machine-size computer that worked with software and add-in cards that were designed for IBM's PC. This was possible not only because the PC was an open system, but also because IBM had used off-the-shelf parts. Any company could buy an Intel CPU and a Microsoft operating system.
By 1984, numerous companies were competing in the IBM-compatible market, including Compaq-like start-ups (Columbia, Eagle, Leading Edge) and established manufacturers (ITT, Tandy). But IBM still led the pack. In 1983, Big Blue released the PC/XT, introducing the hard drive as basic equipment (a configuration with a mammoth 10MB disk went for a cool $4995). And in 1984, it upped the ante with the PC AT, the first PC built around Intel's 80286 processor, running at a blazing 6 MHz.
IBM's lead slipped in 1986, when Compaq shipped the first PC based on Intel's 32-bit 80386 (or 386 for short). As a chip, the 386 was a landmark that made today's windowing, multitasking environments possible. Overall, though, Compaq's system was little more than an AT clone with a better CPU and faster RAM access. Still, it was the latest, greatest PC of the time--and it wasn't from IBM.
Big Blue's dominance faded further in 1987, when the market failed to
accept its much-hyped PS/2, the machine designed to replace the AT. Suddenly,
And they still ran MS-DOS. In fact, most of them are
Not that anyone would have compared Windows 1.0--which was announced in 1983 and shipped in 1985--to a mansion. In fact, this rudimentary graphical front end for DOS was widely derided as clunky. Besides, Microsoft and IBM were soon drumming up enthusiasm for a would-be DOS successor known as OS/2. But when OS/2 finally shipped in 1987, it was tough to configure and slow, and it attracted only a cult following.
In 1990, IBM and Microsoft went their separate ways. IBM continued to improve and sell OS/2, and Microsoft bet the farm on Windows. And that same year, Windows 3.0 changed everything. A huge advance over earlier Windows versions, it became the first environment other than DOS to come preinstalled on most PCs.
Like the original IBM PC, Windows benefited from a wide range of apps.
According to Jeff Tarter, editor and publisher of the industry newsletter
No company entered the Windows application market as aggressively as Microsoft itself. As the folks in Redmond (Microsoft's home since 1986) introduced one product after another, they seemed to leave less and less room for others. It didn't help matters that 1-2-3, WordPerfect, Harvard Graphics, and other key DOS programs moved to Windows only after counterparts from Microsoft had gained a toehold.
"In the mid-1990s, it was very difficult to compete with Microsoft," contends Philippe Kahn, whose Borland International was a major force in software in the 1980s and early 1990s. "As a consequence, competitors have disappeared and products haven't evolved as much as they did in the 1980s."
In 1993, Microsoft released Windows NT, a 32-bit version that was truly an operating system of its own--it didn't require DOS at all. Meant for networks and high-end users, NT traded ease of use for administrator controls and security. Stiff hardware requirements and compatibility issues kept it out of the mainstream.
Two years later, Windows 95 shipped to far greater fanfare. It too was a 32-bit environment that made full use of the latest processors. But DOS was still there, loading before Windows. Microsoft has kept that same basic setup for Windows 98 and Me; when Windows XP arrives, probably this year, home users will get their first DOS-free version of Windows.
With or without DOS, Windows may simply be too complex to be perfectly reliable, with too much old code on top of new code. "Twice as many lines of code probably means four times as many bugs," warns Kahn.
The sheer versatility of the PC and Windows is another source of trouble: The huge number of apps, peripherals, add-ins, and configurations makes thorough debugging impossible. Other platforms such as Linux and the Macintosh, which are often praised for superior stability, don't offer anywhere near the flexibility of a Windows PC. And they have not had anywhere near its success. According to Bricklin, "People voted with their pocketbooks"--and the PC won because it was capable of doing so many things.
So will Windows and PCs be around forever? Already, it's almost a cliché to say that the standard will dwindle as the Internet comes to dominate our computing experience. Low-cost devices like PDAs and Internet appliances could indeed eventually edge out the PC for taking care of simple tasks such as Web browsing and e-mail.
For the foreseeable future, though, if you want one device that can handle everything from managing a business to playing the latest games, you'll probably still need the flexibility of a personal computer. And even as PCs continue to morph, the basic value of backward compatibility should ensure that IBM's 20-year-old standard will live on. The soul of the new machine, it appears, will remain the brains of an old one.
Major events that have shaped the PC platform--plus some famous flops.