While IBM and Microsoft were involved in their negotiations, rumors of the new PC OS began to float throughout the industry. For a while it was called CP/DOS, or DOS 286, or DOS 5.0, and then finally Presentation Manager. It would have multitasking and multi-threading (whatever that was) and semaphores and all manner of good stuff, but most important, it would have a GUI, just like the Mac! This interface would be based on Windows, only much better, obviously, and Windows and the new interface would be so similar that anyone who developed a Windows product could port it to Presentation Manager when it was ready to ship with a snap of the fingers and a twinkle of a compiler. Write two products for the price of one. And boy, that sounded really good to all the developers!
And though no one would actually confirm the date on which the new wonder OS would ship, everyone assumed it would be sometime in 1986 or at the very least 1987. And that sounded good, too, because by 1986, Atari ST owners had a pretty sophisticated GUI for their machines, for God’s sake, whereas PC types still had to clunk along in character-based DOS. And PC owners were sure getting tired of all those Mac snobs laughing at them and twirling those damned mice under their noses and getting all the girls because Macs were so cool. In fact, a fair number of them started buying up Macs so they could twirl mice and be cool, too. But most were still content to wait for IBM to ship a cool Mac-like OS so that they could twirl their mice while avoiding paying Apple those 50 percent profit margins it got on its systems. But they were sure eager to get their hands on that new OS and those mice.
Then IBM threw SAA into the mix and everything changed.
SAA, which stood for Systems Application Architecture, was an attempt by IBM to develop a cross-platform OS (or something close to it) that would run on all IBM mainframes, minicomputers, and PCs. (An inadvertently hilarious book about this heroic effort was written several years ago. It was referred to by industry wags as “The Soul of a Giant Three-Ring Binder.”) This initiative had been sparked by IBM’s annoyance at having to listen to its archrival in the lucrative minicomputer market, DEC, trumpet that it, DEC, “had it now.”
What DEC had “now” was a unified OS and application environment for its entire product line. In theory, no matter what size computer you bought from DEC, they all used the same OS and ran the same software. (This wasn’t entirely true in practice, but DEC certainly was way ahead of IBM in this regard.) By contrast, IBM supported over a dozen incompatible hardware and OS platforms. Moving an accounting package from, for example, an IBM minicomputer system to a mainframe required an extensive rewrite of the software.
SAA was designed to close this perceived competitive gap, but IBM was targeting a chimera. True, in the late 1980s, DEC’s profits and revenue presented the picture of a company in the pink, but this was an illusion. In reality, DEC’s appearance was more akin to the hectic flush a consumptive develops before death. DEC’s business model consisted of selling minicomputers and small mainframes to companies at the departmental and divisional level. This was precisely the market companies such as Novell and 3Com were targeting with their networking OSs. Herds of Silicon Beasts yoked together with NetWare were a fraction of the cost of DEC’s expensive and overengineered hardware products and required less support. LAN systems were also starting to offer a broader and cheaper selection of software to compete with the minicomputer market’s offerings.
Some ominous reports from the field began to filter into DEC headquarters over defections in the company’s customer base, but these were ignored until it was far too late. Over time, the concept of much cheaper and more choices beat “has it now” hands-down and DEC, along with most of the minicomputer market, disappeared in the late 1990s. For good measure, SAA proved to be a massive waste of IBM’s time and money and eventually sank without a trace as well.
Nonetheless, what IBM said still went, and work on integrating SAA technology into Presentation Manager moved forward. Much of this work involved building support for a whole host of IBM mainframe terminals into the new OS. This all took much time and effort, and it was soon apparent there would be no wonder OS in 1986. In fact, the new OS with the cool graphical interface would not be ready until 1988. Oh, and you know all that stuff about a simple recompile being all you needed to do to port your Windows product to Presentation Manager? Forget it. You were going to have to do a major code rewrite to get your product to run under the new OS after all. Which, by the way, was going to be called OS/2. A nice fit, IBM thought, with its new PS/2 line of microcomputers.
Then, in an act of supreme stupidity that would characterize IBM’s marketing of OS/2 for the rest of the product’s ill-starred existence, the company announced it would indeed ship the first version of OS/2 in 1987. Only it wouldn’t have a cool Mac-like GUI—just the same DOS-like character interface everyone was heartily sick of. Few cared that underneath the hood of the new OS was a quantum-leap improvement over DOS in functionality. With this single stroke, IBM had created TopView II.
IBM’s motive for this act wasn’t hard to discern. OS/2 had become a draftee in the company’s war on the hardware cloners, and it had been assigned to ride shotgun alongside its new computers into battle. By 1987, the company had woken up to the consequences of unleashing the Silicon Beast on the market and was looking to take it all back. The PS/2 line would ship with a new bus, the Micro Channel, that had more patents stuck to it than bugs on a fly strip hanging from the ceiling of a Texas gas station. The new BIOS chip, called Advanced BIOS or ABIOS, reared up and bit you on the finger if you tried to reverse engineer it. The units shipped in April 1987 with plain old DOS, and IBM badly wanted something that could better showcase their new darlings. OS/2 1.0 was it.
While this was all well and good for IBM, software publishers were less than thrilled. Companies were being asked to throw a considerable amount of time and money into supporting an OS version whose sales prospects were dubious. Making everyone feel worse was IBM’s pricing of OS/2 1.0: $340.00 for a retail copy, a price that generated sticker shock. IBM had established a low price point for desktop OSs with the introduction in 1981 of DOS 1.0 for $40.00, and no one thought the OS/2 pricing strategy was a smart move. Once a market’s pricing structure is established, it takes time and effort (and perhaps a helpful monopoly) to change it, if you ever do. Yes, many people would eventually obtain the product via bundling, but strong retail sales would help kick start acceptance of OS/2 and generate sales of OS/2-specific products. And that was unlikely to happen with a $340.00 desktop OS that lacked a GUI.
And speaking of pricing, IBM and Microsoft had placed a $3,000.00 price tag on the OS/2 software development kit (SDK). That was no problem for larger software companies, but smaller firms complained bitterly. Microsoft practically gave away its Windows development tools. Even Apple set more reasonable prices for its SDKs.
IBM also seemed oblivious to the need to provide marketing assistance to independent software vendors (ISVs) building OS/2 applications. The company had no direct mail programs a third party could access that would help promote new OS/2 products. IBM had no expertise or influence in software distribution channels and seemed uninterested in developing any. IBM made no attempt to garner critical “shelf space” in major resellers. There were no co-op advertising programs. There were only a few scattered attempts to build a supporting infrastructure of books, publications, shows, and events that would stimulate interest in buying OS/2 and OS/2-related products. IBM’s attitude was that what had worked for the company since the Great Depression would work today. And, to an extent, it did. Several major publishers, including Ashton-Tate, Lotus, SPC, and MicroPro, as well as a few daring start-ups, committed themselves heart and soul to OS/2.
Exacerbating all the aforementioned issues was an event beyond IBM’s control: a rapid spike in memory prices during the OS/2 introduction. This was a big problem as OS/2 required a “whopping” 4MB of memory to be useful, 8MB to step along smartly, and 16MB to really hum. A 1MB memory stick that was projected to sell for about $100.00 shot up to almost $400.00 before the bubble burst.
What wasn’t beyond IBM’s control was the fact that the company was one of the largest producers of memory in the world at the time and in a position to take advantage of a rare opportunity to use hardware to drive software sales. As a glum product manager from DeScribe, a startup that was introducing an OS/2-specific word processor, pointed out, “OS/2 without memory was a $1,000.00 upgrade. Bundled with a handsomely discounted 4MB memory stick, it was a million-copy seller.” But IBM was unresponsive to that idea.
The name “OS/2″ also proved to be a problem. Many people assumed the new OS ran only on IBM’s PS/2 computers, a misperception IBM did little to dispel. Nor did the existence of two versions of the product, the “standard” and “extended” editions, help matters.
All these factors combined to ensure the introduction of OS/2 1.0 was an unmitigated flop. No one bought the package and no one made any money developing software for it. The desktop market as a whole was becoming restive and showed signs of slipping from IBM’s control. Its PS/2 line met stiff resistance from competitors such as Compaq, which spearheaded an effort to develop an independent hardware platform, EISA. After examining IBM’s stiff licensing and royalty demands, most potential OEMs decided to stay with the existing PC architecture and refused to build PS/2 clones.
Still, this was IBM after all. It set the standards. The software industry turned its impatient eyes toward OS/2 1.1, the “real” OS/2, the one that would finally ship with that cool Mac-like interface. After due deliberation, IBM announced it would ship OS/2 1.1 in October 1988. Considering that by then Mac users would have been using a modern GUI for 4 years while PC users labored in computing’s version of the Stone Age, this seemed rather tardy, but OK. Most people were still ready to wait, though sales of Macs continued to grow briskly.
And while you were biding your time there was this increasingly interesting Windows alternative. Windows 2.0 had shipped in April 1987, and though the critics still mocked it, 2.0 was clearly an improvement over the last version. There was even a special version, Windows 386, that took advantage of some of that chip’s special features. You could do some real work with Windows, especially the 386 version, and now there were even some good software packages for it. Desktop publisher PageMaker, for one. And Microsoft’s new spreadsheet, Excel, which had received glowing reviews upon release. When you bought the package, which came in well under $100.00, it said it had Presentation Manager. In other words, you were sort of getting a sneak preview of IBM’s new wonder OS. That was a nice little bonus, when you stopped to think about it.
Microsoft had also done something really quite clever. It had released a runtime version of Windows to developers, ensuring that if you didn’t have Windows, you could still buy a Windows application you could run on your system. If you were a PC user with a 386 and Windows, you might not be ready to twirl that mouse, but you were certainly entitled to swing it a bit. Windows 2.0 sales started to become quite robust, hitting about 10,000 units a month through the retail channels. But no one got carried away. OS/2 1.1 was on the way.
IBM kept its promise. OS/2 1.1 with the Presentation Manager GUI was officially released in October 1988, on Halloween. It had a cool Mac-like GUI, though by now GUIs weren’t really that cool anymore, just necessary if you wanted to be competitive. The market sighed in relief. There was a flurry of initial purchases. OS/2 appeared ready to take off!
And then everyone found out you…couldn’t…print…with OS/2 1.1.
This was because in addition to providing no support for third-party software developers, IBM had also made no attempt to garner support for manufacturers of non-IBM hardware. And although IBM made some very nice printers, most people hadn’t bought them. They’d bought a wide variety of different printers from different manufacturers, most notably HP LaserJets, and OS/2 1.1 had no idea of what to do with them.
And, unfortunately for IBM, memory prices had remained high as well. Because OS/2 1.1 needed even more memory than OS/2 1.0, upgrade costs were around $2,000.00 per PC. And no, IBM hadn’t changed its mind about a hardware/software bundle of memory and OS/2.
The company then proceeded to make the day of OS/2 developers everywhere by announcing that it had licensed the NeXTStep interface from Steve Jobs’s NeXT Software. Rumors immediately began to spread that NeXTStep would become a part of OS/2.
NeXTStep ran on the NeXT computer, Steve Jobs’s incredibly cool black cube desktop PC that cost $10,000.00 per unit. At that price, no one actually intended to buy a NeXT box themselves, but everyone hoped someone would buy one for them so that they could put it on their desk and look as cool as Steve Jobs. The NeXTStep interface was certainly state-of-the-art, with chiseled icons and slick graphics—no one had ever doubted Steve Jobs’s ability to create great-looking icons. But if it were true that it was going to eventually replace Presentation Manager, why write applications for OS/2 now? Everything would have to be extensively rewritten once NeXTStep was integrated into OS/2. Better to wait. On the other hand, everyone was just sick to death of character-based interfaces. (IBM never did anything with NeXTStep.)
The market lowed and shifted about restlessly. More people went out and bought Macs. Had Apple not been in its way every bit as stupid as IBM, the company was in a position to become the Microsoft of OSs. But we all know how that turned out.
Sales of Windows 2.0 hit 50,000 units per month.
OS/2 1.2 shipped a few months later. Most people still couldn’t print with it. IBM announced it was now talking to a company called Metaphor about its really cool OS and interface. (IBM never did anything with Metaphor.)
The lows and the bleatings became louder. The tension rose higher. It was now 1990.
Microsoft announced Windows 3.0. It looked pretty good. It was inexpensive. It supported 16MB of memory but ran OK in 4MB. The memory bubble had burst. Windows really needed a mouse. There were several good programs available for it, including spreadsheets and word processors, the most popular applications. You could print with it. It was 6 damn years after the Mac had first shipped.
The market bellowed loudly and stampeded toward Windows 3.0.
IBM shipped OS/2 1.3. It could—usually—print. It was highly functional. IBM’s Desktop Software division even had a lineup of nice OS/2-specific applications available for it. When did it ship? It doesn’t matter. No one cared.
Before the herd broke, IBM had one last chance to stop the stampede. In 1988, the company had formed its Desktop Software division in Milford, Connecticut. The group was deliberately staffed with young honchos from outside the company who were supposed to show IBM how to succeed in the rough-and-tumble world of PC software. In short order, Desktop Software built a fairly polished stable of OS/2 applications, including word processing, business presentation, and desktop publishing products. All the programs were scheduled to be available in Windows versions, though these were going to ship well after their OS/2 counterparts were out the door.
Alarmed at the growing presence of Windows, and aware that OS/2 needed more time to build momentum in the marketplace, the Desktop Software group petitioned to meet with no less an august personage than IBM President John Akers himself to explain the situation. They had taken a close look at Windows and, despite Microsoft’s soothing words, realized it presented a serious competitive challenge to OS/2. They were also aware that the development market was on the cusp; if events broke the wrong way, software publishers might be forced to abandon OS/2 if they felt Windows would allow them to meet the pent-up demand for GUI-type products on the PC. After submitting their request, they were duly granted an audience before the Big Blue Throne.
At this point, IBM still had the ability to checkmate Microsoft’s plans for Windows. One way was to buy a new OS from a company called GeoWorks. The company had developed a highly optimized product with a slick GUI that could run in a small hardware footprint; GeoWorks ran with amazing alacrity even on the original IBM PC. This was the path favored by the Desktop Software division.
Another option, one widely discussed within IBM and Microsoft, was to release a version of DOS with the Presentation Manager interface. And, as it had been since 1981, Digital Research was still sniffing about forlornly while proffering GEM and DR DOS. If all else failed, IBM’s final option was to simply threaten Microsoft with termination of the joint development agreement between the two companies and strike out on its own. At this point, IBM still held the upper hand in the relationship and in the marketplace, and Microsoft would have had to back down.
The day of the meeting arrived and the Desktop Software contingent, led by Product Marketing Manager for IBM Corporation, Randy Hujar, was escorted before IBM’s reigning monarch and given the chance to make their case. Akers received a detailed briefing on the situation, as well as a series of recommendations. When the team was done, he called them “a group of good kids” and proceeded to explain the facts of life to the naïfs before him. IBM, he told them, controlled these markets and set the standards, and it always would. Bill Gates “was a nice boy,” and IBM fully understood how to position OS/2 vis-à-vis Windows. It was all well taken care of. They could go back to their cubicles and not worry their precious little heads about the problem.
The Desktop Software group was then escorted out of the august presence and reoccupied their cubicles. IBM proceeded to develop a ludicrous agreement with Microsoft that said that Windows was just great for “low-end machines” (i.e., the ones that most people had) and that OS/2 was great for “high-end machines” (i.e., the ones they would one day own). Windows 3.0 shipped as planned.
IBM shut down its Desktop Software group in 1992, just in time to ensure that the division’s applications wouldn’t be available to support the rollout of OS/2 2.0. That same year John Akers was kicked out of the CEO position at IBM. Microsoft was estimated to have shipped approximately 30 million Windows 3.0 and 3.1 licenses by that time.