November 03, 2016
The accidental standard (2)
When Gary Kildall was interviewed for the third issue of PC Magazine (Jun/Jul 1982), the future of the PC operating system was very much up in the air. The tech press was hedging its bets. CP/M was still the most popular microcomputer OS. (In the mid-1980s, I was using a Kaypro II running CP/M.) Apple, as always, lived in its own proprietary world.
It only took a year for that uncertainty to fall away and things to gel. IBM made the microcomputer respectable and Microsoft made developing applications for other operating systems unnecessary. But they still had to be individually tweaked to account for each manufacturer's BIOS chip and hardware specs.
In November 1982, Compaq debuted a personal computer with a reverse-engineered BIOS, making it truly "IBM-compatible." Eighteen months later, Phoenix Technologies produced its own 100-percent IBM-compatible BIOS chips and sold them to anyone willing to pay the licensing fee (that additionally indemnified its customers from getting sued by IBM for IP infringement).
Microsoft was already selling MS-DOS to all takers and IBM "lookalikes" were flooding the market. But now the era of the 100 percent compatible IBM "clone" had arrived. The market solidified. In the August 1983 issues of PC Magazine, Todd Katz asked, "Is CP/M Dead?"
Naturally, Digital Research product manager Kevin Wandryk didn't think so.
Even if we do lose this marketplace and it goes totally to Microsoft, this is only Round Two. There is the 68000, the Intel 8286 and 8386 [80286 and 80386], the National Semiconductor 16032, and we have a leading edge at the present time in the operating system development in each of those areas. We certainly won't be blindsided again.
They wouldn't be blindsided because the 1970s and its tossed salad of 8-bit CPUs was over. Apple would go with the 68000. Microsoft and IBM and Intel would stick with MS-DOS and the x86 platform. Nobody was pining for alternatives. Katz saw the writing on the wall. His answer: "CP/M-86 is worse than dead, it is irrelevant."
In the October 1983 issue of PC Magazine, Compaq chairman Benjamin Rosen prophetically predicted that three players would remain in the market: "Those adhering to [the IBM PC] standard and those named Apple" and everybody else.
What came to be known as the "Wintel" standard (Windows + Intel) mattered so much that even the "IBM" part faded to insignificance. Compaq upped the game in 1987 with the Deskpro 386, the first PC to run on the 32-bit Intel 80386 chip. An IBM-compatible that was more "compatible" than an IBM PC forced IBM to lead, follow, or get out of the way.
IBM decided to lead, attempting to reassert sovereignty over the PC world with OS/2. OS/2 and the proprietary Micro Channel bus would lock users into the IBM ecosystem. In its struggle for market share, OS/2 was touted as "a better DOS than DOS and a better Windows than Windows," and that was the whole problem. Everybody was happy using DOS and Windows.
The same way nobody had wanted or needed CP/M-86 once MS-DOS had established itself among vendors and users, nobody wanted or needed yet another x86 OS standard. And Microsoft, who had developed OS/2 with IBM, quickly decided that it didn't either.
IBM and Microsoft broke up in 1990. Microsoft said it was sorry with a billion dollar alimony payment. Back in 1988, Microsoft had hired VMS architect Dave Cutler (another connection between DEC and Microsoft) to create NT, its multitasking "protected mode" OS. By the release of Windows XP in 2001, NT had turned into "a better Windows than Windows" that was still Windows.
|Courtesy Wikipedia Commons.|
Microsoft has never forgotten that lesson, only abandoning native 16-bit MS-DOS (DOS!!!) compatibility with the shift to 64-bit processing. I still use an old WordPerfect DOS dictionary app on my Windows XP machine, and all my 32-bit Windows 95 apps run just fine.
Well, Microsoft did forget it temporarily with Windows 8, when it pretended to be Apple. Apple, remember, had pulled the rug out from under its user and developer base at least three times: switching from MOS Technology 6502 to Motorola 68000 to PowerPC to Intel x86 CPUs.
Microsoft only messed with the interface of Windows and was forced to beat a hasty retreat. Though let's not forget that Windows still owns 90 percent of the (albeit shrinking) desktop/laptop market.
Microsoft getting ahead of itself with Windows 8 was a consequence of
it getting behind the curve with the Windows Phone. And that takes us way back to the beginning and Digital Research's late arrival to the PC party with CP/M-86.
CP/M (like DOS 1.0) wasn't a "standard," but Digital Research was open to customizing the operating system for every 8-bit CPU that came down the pike in the 1970s. The resulting fragmentation and version control problems meant that computers in the same product line often weren't compatible with each other, to say nothing of competing platforms.
Along with Palm and Blackberry, Microsoft was an early player in the mobile OS market, developing Windows CE since 1996. Like Digital Research, it had gotten good at customizing Windows CE for each vendor in a heterogeneous hardware market.
As Digital Research was by IBM and MS-DOS, Microsoft was blindsided by the iPhone, build on a single hardware platform with a new interface. Then Google made Android the DOS of the smartphone, licensing it to all comers. Google could have cribbed from Microsoft's famous mission statement: "A smartphone in every pocket all running Android software."
Back in 1983 and 1984, industry prognosticators were predicting that, any day now, MS-DOS would be superseded by Digital Research's CP/M-86, or Xenix (Microsoft's version of UNIX), or PC/IX (IBM's version of UNIX). A few years later, OS/2 and Micro Channel were going to dominate for sure.
But the IBM PC had set the standard and the PC world didn't need or want another one. Not even IBM could alter the ultimate direction of its own creation. This explains Microsoft's draconian efforts to get old fuddy-duddy hold-outs (like me) onto Windows 10: fragmentation and loss of version control is death.
The evolution of the PC made clear that the consumer market has room for two operating system (Windows and Mac), with the third (Linux) ending up a couple of sigma out on the long tail. The same thing happened with Android and iOS in remarkably similar proportions, this time with Windows Phone ending up with single digits of market share.
Unlike Digital Research, Microsoft has the resources to stay in the race. It plans to focus its Windows Phone efforts on enterprise customers while "betting on a technology leap in a few years with a paradigm shift." Which I take to mean: when Continuum and a Surface phone become practical realities (and Apple loses interest in the desktop OS).
Considering how much the computer industry has changed in the past 35 years, I won't be surprised at all if and when a new "accidental standard" takes over in a flash.