| Back to Dec 95 First Impressions | Up to Table of Contents | Ahead to Dec 1995 Features |
by: Fred Langa, editorial director, Robert Lauriston, contributing editor,
and Diganta Majumder, news editor
Click Here to see a
21.0KB bitmap image of artwork
which goes with this article, entitled:
America Goes High-Tech
When Microsoft first announced Windows in November 1983, the typical PC looked something like this: 4.77MHz 8088 CPU, 64KB of RAM, character-mode graphics and a 320KB floppy drive--or maybe two drives if you were lucky.
Despite the paucity of technology on the desktop, there was already a broad consensus about the future of computing--and that vision demanded much more than was available at the time. Like the Xerox Star and the Apple Lisa, the PCs of the future would have bitmapped displays and mouse-driven interfaces that could display multiple applications in separate windows. Like UNIX, they'd be able to multitask and would use gobs of RAM. Given what we know today about GUI hardware requirements, this early vision was almost science fiction, but that didn't faze the heavy hitters in their battle to set the standard.
The GUI era began in earnest in late 1982, when VisiCorp announced VisiOn, a windowing environment that would run multiple applications on top of DOS. Over the next two years, many similar announcements followed from a variety of industry leaders such as Digital Research, IBM and Microsoft, as well as from startups such as Quarterdeck and Trillian. The products, slow in coming, popularized the term "vaporware."
Meanwhile, Microsoft kept pushing back Windows' April 1984 ship date. Though the first software developers' kits went out in May 1984, the retail version didn't ship until November 1985. By way of apology, Microsoft turned the rollout into a Friars Club-style roast, with industry pundits making fun of the company's executives and Stewart Alsop giving Bill Gates a "Golden Vaporware" award. And Windows 1.0 itself? Not bad for a first try, but its RAM and disk requirements guaranteed that the new OS would languish on store shelves.
Shortly before the Windows rollout, IBM and Microsoft announced they were working together on future PC operating systems. This was a big relief to Compaq and the other cloners, since it meant they would still be able to sell computers that ran the same software as IBM's machines. Microsoft was relieved, too, as it had feared IBM might look elsewhere for an OS for its rumored "PC-2" line.
Microsoft's main goals for the new OS, eventually named OS/2, were to break the 640KB barrier (just then becoming a familiar obstacle for power users) and to add multitasking. The new 286 CPU addressed both objectives--sort of. Its "protected" mode could run programs as large as 16MB. DOS apps couldn't run in that mode, but they could run in the 286's "real" mode, which emulated the 8088 CPU. However, thanks to some unfortunate design decisions made by Intel, switching between those two modes was very difficult. As a result, the 286-based OS/2 could not multitask DOS applications.
To benefit from OS/2, in fact, you would have had to replace all of your apps. Microsoft hoped users would accept that inconvenience and expense for the chance to run larger apps. But then the DOS extenders arrived, making it possible to write larger apps without abandoning DOS. Meanwhile, clone vendors released PCs based on the 386 chip, which had no problem switching between protected and real modes. By the middle of 1986, Quarterdeck was shipping a 386 version of DesqView that could multitask regular DOS apps and large DOS-extended apps as well.
Microsoft wanted to start over and make the first joint release a 386-only OS, but IBM (still without 386 systems) insisted on going forward with the 286-based OS/2. The two companies were also arguing about OS/2's graphical interface: Microsoft wanted to use Windows, while Big Blue pushed its own code. The issue was still unresolved when the two companies announced OS/2 in April 1987. IBM discussed "a presentation manager" and emphasized compatibility with its minis and mainframes. Meanwhile, Microsoft talked up the Microsoft Windows Presentation Manager and pointed out how the forthcoming Windows 2.0 would have the same interface.
Though Microsoft had moved Windows to its back burner, that market was slowly heating up. In October, Microsoft finally shipped the first major Windows app, Excel, to good reviews and fast-growing sales. In December, it released Windows/386. Although the OS didn't break the 640KB ceiling, it could multitask as many DOS and Windows apps as would fit in memory. Compaq and other 386 vendors started bundling it with their hardware, and while most of those copies ended up as shelfware, the environment attracted a following among in-house corporate developers.
Meanwhile, OS/2 was going nowhere. The character-mode-only 1.0 version released at the end of 1987 had performance and compatibility problems, particularly with non-IBM hardware and apps that used undocumented APIs (including Windows). The full-GUI 1.1 release had the same limitations, and for decent performance you needed a 386 with at least 6MB of RAM--quite impractical at a time when the typical new PC was a 640KB 286 and RAM upgrades cost around $150 a megabyte.
Nevertheless, OS/2's eventual dominance seemed inevitable. Its future seemed so secure that the only questions bandied about by analysts and the trade press were how long it would take before OS/2 supplanted DOS, and whether UNIX might be able to gain a share of the market.
Market leaders Lotus, WordPerfect and Ashton-Tate (then number-one in spreadsheets, word processors and databases) were all hard at work on Presentation Manager apps and had no interest in writing anything for Windows. Why should they write for APIs controlled by Microsoft, one of their most determined and capable competitors, when they could use a look-alike controlled by IBM, never much of a player in their markets? Moreover, if Apple won its look-and-feel lawsuit against Microsoft, it might force Microsoft to withdraw Windows, killing the market for Windows programs.
Back at Microsoft in late 1988, the Windows team had a rough version of Windows 3.0, which met the two goals Microsoft had set for OS/2: It broke the 640KB barrier, and it could multitask both Windows and DOS apps. As the betas started getting around in 1989, OS/2's future no longer looked so secure. Word got back to IBM, which started work on PM Lite, a Windows-like environment that would run the Presentation Manager APIs on a DOS foundation.
Microsoft managed to fast-talk IBM out of doing PM Lite, and the eventual compromise was announced in November. IBM endorsed Windows as the environment for systems with less than 2MB of RAM, Microsoft reiterated its position that OS/2 and PM were the platform of the future, and the two companies pledged to reduce OS/2's minimum RAM requirement from 3MB to 2MB. But 2MB was barely enough to run Windows 3.0, and OS/2 needed at least twice that. The resultant joint press release was full of incomprehensible double-talk, and astute observers wondered what was going on. In retrospect, it seems clear that both companies were stalling for time, waiting to see the market's response to Windows 3.0.
Microsoft released Windows 3.0 in 1990 amid an unprecedented promotional blitz. The program was a significant improvement over Windows/386, but its immediate success was due more to good timing. With the 386SX/20 driving the 286 to extinction, even entry-level systems had the horsepower needed to run Windows 3.0. Over the next two years, Microsoft would ship, by some estimates, nearly 10 million copies. But Windows 3.0 still had some rough edges and it crashed a lot, so most of those ended up as shelfware, too. It's a safe guess that in early 1992, only about 3 million people were running mostly Windows applications.
IBM and Microsoft started divorce proceedings shortly after Windows 3.0 came out. IBM took OS/2 2.0 back from Microsoft and started working on adapting it to run Windows apps. Big Blue eventually succeeded well enough that developers had less incentive than ever to write native OS/2 programs--but not well enough to lure many users away from the real thing. Microsoft kept its 386-only successor, OS/2 3.0, which became Windows NT.
Windows entered its "everywhere" phase in 1992, with the release of version 3.1. After the 486 killed the 386, even the cheapest systems ran Windows fast, and Windows versions of almost every app were readily available. Proof that Windows was shelfware no more came in the first quarter of 1993, when the Software Publishers Association reported that Windows apps outsold DOS apps, $669 million to $502 million.
Not much has changed since then. Windows ships with almost every PC sold, and as people replace their old XT and 286 machines, almost all of them switch to Windows. With total domination of the market, the only way for Microsoft to increase sales is to make the market larger--hence Windows 95's focus on appealing to people who've never used computers. Fifty million users may sound like a lot, but Microsoft still has another 5 billion users to go before it can truly put "Windows everywhere." --RL
Robert Lauriston is a contributing editor for WINDOWS Magazine and co-author of The PC Bible (Peachpit Press, 2nd ed. 1995), winner of the 1994 Computer Press Award for Best Introductory How-To Book: Systems. To find his E-Mail ID Click Here
Many critics of Windows--and even some of its supporters--have long had one big question about the operating system's success: How? How did it squeeze out technically comparable, perhaps superior, rivals like Apple's Macintosh and IBM's OS/2?
Apple could have been Microsoft, says Karl Wong, principal analyst for PC software at research firm Dataquest. A decade ago, things seemed to be going that way. Remember the one-time-only ad Apple ran during the 1984 Super Bowl to launch the Macintosh? With an adroit mix of Orwellian doom and wicked humor, the spot poked fun at the big-iron universe of IBM and forever changed the image of personal computing, introducing a hip new world in which computers were cool and great power could be had right from the desktop.
But then Apple stumbled--badly. It refused to license the Macintosh, fearful that any dilution of its proprietary status would lead to a loss of quality. What it lost instead were sales. Since Apple was the only place you could get the hardware (except for add-in cards), the Mac's price stayed high; some PC clones could be had for half as much.
Microsoft, by contrast, understood early on the business of cloning, and it willingly licensed DOS to any hardware maker who wanted it. That not only gave the company market share early in the game but money in the bank to live on while Windows built a market for itself.
Moreover, while Apple earned critical raves for its cutting-edge innovation, Microsoft developed, manufactured and marketed Windows as a consumer product instead of gee-whiz technology. To Microsoft, its operating system and applications alike were designed not to be marveled at but rather to solve users' problems. The company eschewed glamour and went for mass appeal. Over the years, that was how it played out: The Macintosh got the intense user loyalty, and Windows got the sales.
The story behind OS/2 was more convoluted. Microsoft and IBM co-developed the early versions, which, like every product in the category, had a very slow start. The first incarnation, released in December 1987, was text-only; when the GUI did arrive, it was clearly inferior to the Mac. While Microsoft engineers continued to work on subsequent versions of the product, the company was convinced its independently developed Windows was the way to go.
As the rift between the two companies got ever wider, each formally went its own way. Not surprisingly, the market lined up behind IBM; conventional wisdom had it that once OS/2 finally got its act together, it would be invincible.
But when Windows 3.0 arrived in mid-1990, the timing could not have been better. The operating system was hardware-independent, ran on 4MB 386 machines, and had the ability to multitask DOS programs. Within a month, it had outsold all versions of OS/2 combined, and software developers couldn't defect fast enough. And it didn't hurt that Microsoft, which had built lucrative licensing deals for DOS with virtually every OEM, put those relationships to very good use to propagate Windows. (In the process, it earned itself a reputation for hardball tactics that persists to this day.)
Today, Mac devotees can't hide their anger at the hype accorded Microsoft's upgrade this summer (cleverest slogan: "Windows 95=Macintosh '89"). And IBM has committed millions of ad dollars to OS/2 Warp since its release last year. But Warp never really had a chance on the desktop, and Big Blue has repositioned it for the server market.
Maybe it's just canny marketing, maybe it's brutal business, maybe it's the result of a deal with the devil. A decade after the battle began, the Mac still has the class, and OS/2 still has IBM's clout. Windows--still looking for respect--just gets the customers. --DM
For good or ill, much of the history of desktop computing hardware has been shaped by Microsoft and its partners. Some companies (such as IBM) flew highest in the desktop realm when they were allied with the Redmond giant, but have since faltered. Others (like Apple) have always tried to march to a different drummer, even though the percentage of people in their parade is shrinking as Microsoft's grows.
Over the past decade, Windows has hugely affected our hardware. Once a curiosity, Windows has reached a point where essentially all new hardware is Windows-compatible, and something like 96 percent of all new systems ship with Windows pre-installed.
Here are some of the hardware highlights of the past decade-plus that mark Windows' path to notoriety.
1983 Apple built on its phenomenal success in character-oriented small computers by rolling out the hot-selling Apple IIe at $1,395. It also tried to break new ground with the Lisa, the first attempt at a mass-market computer with a graphical user interface. Words like "mouse," "icon," "pointer" and "windows" became part of the computing lexicon. The machine generated raves for its advanced technology, but its relatively slow performance and astronomical $10,000 price tag kept it out of the hands of most users.
Meanwhile, sales of the original IBM PC skyrocketed, and IBM rolled out two follow-on machines: a huge loser in the chicklet-keyed, anemic and kludgy PCjr; and a huge winner in the $4,995 IBM XT, which added a then-mammoth 10MB hard drive and three more slots to the basic PC design. Like the wildly successful IBM PC, the XT ran PC-DOS, Big Blue's version of Microsoft's system software.
Meanwhile, non-PC-standard computer makers were falling on hard times. For example, Coleco's proprietary Adam computer was met with such indifference that wags soon rechristened it the "Adam bomb." A major shakeout among vendors of non-IBM-compatible machines ensued.
This year was also notable for its first major break from the desktop. Radio Shack introduced the world's first successful laptop computer, the four-pound TRS-80 Model 100. Its great keyboard, built-in modem and 20-hour (!) battery life--using ordinary AA batteries--made it an instant hit. The unit sported a small but clear LCD screen and was primarily character-based.
1984 Having learned from the Lisa, Apple priced the first Macintosh at $2,495. It sported 128KB of RAM. It also introduced millions to the graphical interface and the hard-shell 3.5-inch floppy disk, as well as to unfamiliar terms such as "desktop," "clipboard" and "cut and paste." Almost overnight, the bluish glow of the Mac's 9-inch monochrome monitor (capable of 512x342-pixel bitmaps) could be seen illuminating the cubicles of computing's avant-garde.
Back in the mainstream, IBM rolled out the mighty 16-bit, 286-based AT machine, whose basic system architecture still forms the foundation of much of today's hardware. Fortunately, not much else of the AT survived: A fully decked-out AT came with a 6MHz CPU, 512KB of RAM, a 20MB hard drive, a monochrome monitor, a new 1.2MB floppy drive and PC/MS-DOS 3.0--all for $6,694. Lots of older hardware and software written for the IBM PC would not work properly on the new system.
1985 By 1985, the IBM/Microsoft standard was steamrollering the competition, besting even arguably superior platforms. For example, Commodore delivered an advanced multitasking, graphical operating system in its Amiga line of computers, garnering rabid support from its users. But poor marketing and lack of supporting applications consigned the Amiga to niche markets. It never recovered.
All the activity on the PC side of the industry meant that some of the Mac faithful started to feel left out, and we saw the first intimations that the Mac's lock on graphical computing wouldn't be enough to carry it to general market dominance. One example: Dayna Communications rolled out MacCharlie, a way to run some DOS software on a Mac. Ten years later, Apple and various other companies are still trying to do this.
1986 IBM dipped its toes in the RISC waters with the RT PC, but no one cared except keyboard vendors, who quickly adopted the RT's 101-key keyboard and made it an office standard that survives today.
The IBM AT got goosed to a screaming 8MHz, and a cottage industry was born as hobbyists found ways to increase the clock speeds of their 286-based systems to 10 and even 12MHz. The chips ran hot and died young.
On the cutting edge, the then-unknown Advanced Logic Research rolled out the world's first production 80386-based machine, the Access 386. It was followed by Compaq's first 386, the Deskpro 386. Suddenly, for a lucky few, PC-standard 32-bit computing had arrived--at least for hardware.
1987 Apple retained a clear lead in graphical computing throughout 1987 with its new Mac II and Mac SE. IBM, without a hardware hit since the AT and losing steam, debuted its new line. The PS/2 came with the advanced and proprietary Micro Channel bus. IBM was trying to regain control of the standard it had created but then lost to an army of cloners.
Although IBM hired the stars of the TV hit *M*A*S*H* to hawk its new line, the PS/2 line suffered from lackluster sales. Compaq designed a PS/2 clone, but then announced it wouldn't sell it. Meanwhile, off-the-shelf AT-class clone machines legitimately reached 12MHz clock
speeds, and more and more 386-based systems began to appear. Systems using EGA (Enhanced Graphics Adapter) and color monitors started making a dent in a world filled with green and amber monochrome systems.
1988 Apple alumnus Steve Jobs rolled out his NeXT system--a computer in a sleek black magnesium cube. It was graphical, object-oriented and way cool. But instead of being a real challenge to the Mac, NeXT was aimed at the education market. It used a very unusual writable optical disk for storage. And it cost $10,000. It sold slowly.
On the PC side, system hardware took a half-step backward: Intel rolled out the 386SX, a 32-bit 386 chip in an externally 16-bit package. The idea was to make 386-based systems cheaper. But confused consumers, incorrectly thinking that the newest 386 chip must be the most powerful, bought the SX-based systems in droves. Some SX systems sold for higher prices than their true 32-bit counterparts.
But PC video improved markedly. The first VGA (Video Graphics Array) systems shipped in quantity, heralding the arrival of another standard that survives to this day. And computer writers were drooling over the new crop of "high-speed" 40MB hard drives.
1989 Computer magazines announced "the fastest systems ever" when 33MHz 80386-based systems appeared, and 25MHz 386s started showing up on the desktops of mere mortals. Around the same time, expensive, high-end file-server systems based on Intel's next-generation 80486 chip showed up. Example: ALR's $13,000 486 machine.
Meanwhile, MS-DOS still ran on everything from the tiny, 8088-powered, one-pound Poquet PC to Compaq's SystemPro, the first mass-market multiprocessor machine.
1990 The gulf between IBM and Microsoft reached oceanic proportions. IBM tried to make waves again, this time with the PS/1. But only a few ripples were generated.
The 286 began to slip out from beneath the middle of the computing bell curve, and the 33MHz 386 became the must-have high-end desktop system. Various cloners pushed the envelope with nonstandard 40MHz and 50MHz designs. Prices on 486 systems began to fall out of the stratosphere. Modem users wondered what to do with all the bandwidth in the newly popular 9600-bit-per-second units.
1991 The 486 became the new must-have system among desktop computer users in the upper ranks of their companies' silicon food chain. Normal, everyday hardware was finally powerful enough to do something meaningful with Windows, and as the newly refined Windows was married to powerful, low-cost clone hardware, some of the benefits of formerly expensive Mac-based GUIs were available cheap. The raw numbers of Windows-based systems climbed sharply.
IBM, now tied to OS/2 software and PS/2 hardware, tried to introduce yet another video standard, the XGA (Extended Graphics Array). It was technically slick, but the video market was starting to focus on Windows performance on clone machines. As a result, XGA languished while coprocessor-based video accelerators that boosted Windows' speed were big news. Example: Orchid's Fahrenheit 1280, at $399.
Newly aware of the importance of graphical displays, people paid more attention to monitors than ever before. Color portables began shipping in quantity, among them Toshiba's T5200C, a Windows-capable--but very expensive--machine. With a 20MHz 386 chip, 2MB of RAM, a 200MB hard drive and a 10-inch passive-matrix color screen, it cost $9,499.
Despite the numeric growth of Windows and Windows-associated hardware, percentage-wise it was still an overwhelmingly DOS world. Most ads from major vendors still showed mouseless machines running DOS apps--even when these ads ran in the newly hatched WINDOWS Magazine.
1992 Clock speeds on the 486 rose, offering more and more power at lower and lower costs.
With new muscle at their fingertips, Windows business users began to explore areas once reserved for the Mac. For example, "Windows-based desktop publishing" was no longer an oxymoron. Alternate input devices became increasingly available for Windows systems. Microsoft's Windows Sound System shipped in 1992, as did Turtle Beach's Multisound "with waveform capabilities."
ATI's Graphics Ultra Pro also shipped, starting a lineage that continues today. Double-speed CD-ROM drives arrived, and color portable prices plummeted. An AST passive-matrix color system now cost less than $3,000.
DOS was still king, but Windows was by now an explosively growing phenomenon. System vendors responded by making mice standard (instead of added-cost options) and advertising systems "optimized for Windows."
1993 This was the milestone year when Windows eclipsed DOS. Windows was finally everywhere, and most new systems shipped with Windows pre-loaded.
The initially pricey Pentium arrived, but systems were in short supply due to what Intel called "technical problems." The first of the 486 clone chips shipped, and Intel soon overcame its technical problems, pumping out Pentiums in quantity. Clock-doubled 486s fleshed out the chip choices power users faced. VL-bus systems debuted, and super-fast--but often slightly weird--VL-based video cards arrived to push Windows' graphics performance to new heights.
The CD-ROM hit the mainstream: Everyone and his brother offered double-speed CD-ROM drives, and NEC rolled out a short-lived triple-speed before the industry leapfrogged to quads. Hayes shipped its Optima 144 Data + Fax144, Pocket Edition fax modem (for $519). Omnibook "superportables" arrived to run Windows, and color printers no longer required a second mortgage.
1994 The trade press began speaking of the "Wintel" (Windows and Intel) standard the way people used to speak of the Microsoft/IBM collaboration: Windows on Intel-based hardware (almost no one thought of it as IBM-inspired anymore) was the new standard, creating a new wave.
Pentium speeds increased toward the triple digits. Clock-tripled 486s were available in abundance; 28.8Kbps modems made their debut; half-gigabyte drives became commonplace and the industry began to consolidate on 15-inch monitors instead of 14-inch models. Laptops became more powerful than the desktops of two or three years before. It was almost impossible to buy a system without some kind of decent video acceleration. It was also almost impossible to buy a system without Windows pre-loaded on it.
Today P90s are the emerging desktop standard, with Pentium 133 systems the current high-water mark as we await the arrival of the first 150MHz PCs. The first systems using the Pentium's successor--the P6, or Pentium Pro--are just starting to roll off the assembly lines. Graphics boards routinely offer 64- and even 128-bit power, while the distinction between a portable and a desktop system lies mostly in the size of the case. Full-gigabyte hard drives are standard on new systems, and 17-inch monitors are increasingly common.
Everyone and everything runs Windows. An estimated 96 percent of all new systems ship with Windows pre-loaded, and Windows' overall market share tops 80 percent, with the Mac, OS/2, UNIX and everything else divvying up the other 20 percent. Pundits begin to worry about the scope of Microsoft's dominance: None of Microsoft's competitors--IBM, Apple, Novell, Lotus--seem able to mount an effective counterattack.
There's no sign this will change soon. With Windows 95 on current Intel-based hardware, Windows systems have finally become fully as good as the Mac--better in some areas, a little worse in others. Some observers predict the Mac's demise, except in small niche markets. And IBM says it will not support Win95 applications on OS/2, which almost surely means that the relative drought of top-flight OS/2 applications will continue.
In 10 years, Windows has climbed from shelfware to everywhere. Those who rode the wave with Windows are today's brightest stars. Those who fought the wave--including giants like Apple and IBM--are hurting.
It's a Windows world out there. --FL
Microsoft wasn't the only company determined to bring a GUI to the PC. Here's a quick look at a few GUI also-rans.
Senior vice president of the Personal Systems Division of Microsoft, Brad Silverberg shared his vision of the Windows industry with WINDOWS Magazine executive editor Mike Elgan.
Q: What are the major surprises, either positive or negative, that you (or, collectively, Microsoft) discovered after the launch of Windows 95?
A: Before we shipped, we got a letter from a guy who so badly wanted to be part of the beta program for Windows 95 that he included an X-ray of a recent bullet wound and a note from his doctor explaining why his beta application was late. An isolated fanatic is one thing, but I was amazed to see the people who lined up at midnight to be the first to have Windows 95! I have always said, it's not a dessert topping, it's not a floor wax, it's just Windows ... but I am starting to wonder.
Q: Windows 3.x is still selling briskly. Why?
A: Windows 3.1 isn't really selling at retail, which is understandable given how much interest there has been in Windows 95 and Windows 95 products. But we continue to make Windows 3.1 available to corporate customers who need some time to make the transition. This is what we have always done--in fact, it is still possible to buy Windows 3.0, though I can assure you it is not a hot seller.
Q: Will Windows and NT merge?
A: Windows 95 shares more code with NT than does Windows 3.1--for example, we are borrowing code from NT for our TCP/IP stack, the multitasking scheduler and OLE. NT will use the Windows 95 shell code in a future release. Someday, we'd like to share the same kernel, although it won't happen for a few years and several releases. It's something we can consider when the mainstream hardware platform is more like a 16MB Pentium system, and when customers tell us they no longer require compatibility with old 16-bit drivers and legacy applications. But don't confuse sharing the kernels with the idea that we'll have a one-size-fits-all product. In fact, it is quite likely that you will see separate packages for servers, high-end business desktops and home machines.
Q: If Dustin Hoffman were graduating today, what word would you give him instead of "plastics" as your advice about the coming decade?
A: First, I'd ask him what took so long. Then we might talk about some of the coming revolutions in personal computing, such as the personal computer becoming a tool for the whole family--for traditional tasks, to be sure, but also for learning, entertainment and communicating.