The Rise and Fall of the IBM PC Part 3: The Cloners Strike Back
Welcome to Another Boring Topic. When we left off in part 2 of this series, IBM had just realized that they were rapidly losing control of the PC market, a market that they themselves had created, and were in serious danger of becoming just one of many PC compatible makers, albeit one with much higher costs and significantly slower R&D processes than their competitors. The gold rush mentality that was driving explosive competitive growth in the IBM compatible market was so strong that even companies who had built their business on entirely separate computer systems were starting lines of IBM compatibles, such as Commodore with its unimaginatively named PC series and Tandy with the equally creatively named Tandy line of compatibles.
Commodore mostly kept its focus, insomuch as Commodore’s dysfunctional leadership was capable of keeping a focus on anything, on continuing to sell the Commodore 64, its successors, and its shiny new line of Amiga computers, but there just was too much money out there in the PC compatible gold rush to pass up the opportunity to at least tap into it. But it wasn’t just large and well established companies churning out IBM compatible clones, new upstarts were also mushrooming up, such as Dell, founded out of Michael Dell’s college dorm room. The end result of this was that IBM’s market share of the personal computer market contracted down to 38 percent in 1987, down almost half from its high of 63 percent in 1984. And this is an even more significant figure in light of the fact that the entire personal computer market of 1987 still included major players such as Apple that were not selling PC compatibles. Six years after introducing the original PC, IBM was reduced to merely being a major player in the PC compatible market. And there was no sign of the bleeding stopping anytime soon, with a constant stream of new compatibles coming to market that undersold IBM’s offerings while offering more features.
By the mid 1980s the barriers to entry into the PC compatible market had been made even lower by thanks to the proliferation of OEM Asian companies that manufactured complete IBM compatible white label systems, that could be bought cheaply by a US garage based startup, who would then slap their own logo on them and undercut IBM’s offerings even further. This cannibalization of IBM’s market share stemmed from two major weaknesses in IBM’s control of the PC standard. The first one was the open, off the shelf architecture of the PC. With the exception of the BIOS chip, essentially the entire architecture was available to anybody to copy, specifically the ISA bus. As a side note, for consistency sake I will be referring to the standard PC bus as the ISA bus although it was variously known as the PC/XT bus in its original 8 bit configuration, the I/O Channel once it was updated to 16 bit with the release of the AT, although it was then usually called the AT bus, and then was finally retroactively named the ISA bus by a consortium of the various major clone manufacturers of the late 80s, and that’s the name typically used today. The bus is how the various pieces of a computer, such as disk drives, memory, and the processor communicate with each other. The IBM bus was hardly the first bus to be used on a successful
personal computer, as every personal computer on the market had one, from the Altair bus, more commonly known as the S-100 bus, to the Expansion Interface box that the TRS-80 used. However the ISA bus was the one to meet broad market appeal once the PC clone market took off as compatibility with it meant that a hardware plugin board would work on every single PC compatible. As a reminder, IBM hadn’t really expected this to be a problem since it assumed its proprietary BIOS chip would prevent any cloners from legally making clones of the PC.
And illegal clones could be easily handled by IBM’s massive phalanx of lawyers. Sadly, the BIOS chip had not proven the obstacle to cloning that IBM had assumed it would be. It was bad enough when Compaq reverse engineered it for its own products, but at least they kept it for their own clones and didn’t sell it to others. The worst was when Phoenix Technologies reverse engineered the BIOS for sale to anybody that wanted to build a clone without the expensive and tedious process of reverse engineering the BIOS chip themselves. IBM had also painfully discovered that both Compaq and Phoenix Technologies’ cloned BIOS copies were one hundred percent legal proof. But this was just half of IBM’s problem. The second major problem was the deal IBM had made
with a certain rapidly growing company based out of Seattle. Microsoft’s youthful leader, Bill Gates, had successfully licensed MS-DOS to IBM for the original PC, without selling the rights to it. This enabled Microsoft to sell MS-DOS to any other computer company that was interested in purchasing it, and as previously mentioned, there were many of them. Gates and
Microsoft were making a fortune licensing MS-DOS to every company that came calling. Furthermore, Microsoft had a very strict policy that under no circumstances would they ever lose out on a sale. They would do whatever it took to get a would-be cloner to sign on the dotted line, prices were always negotiable and one Microsoft employee remembers that there really was never even a price list published for MS-DOS as every deal was separately negotiated.
A further annoyance to IBM was the fact that the DOS operating system they had licensed almost always had the letters “MS” prior to it when people talked about it, and their attempts to get people to refer to it as “PC-DOS” instead had been unsuccessful. IBM had to solve both problems simultaneously in order to regain control of the PC standard. They needed a new hardware solution that was custom to IBM and protected by as many patents, contracts and drooling lawyers as it took to fully lock it down from unauthorized cloning. And second, they needed to come up with an operating system replacement for MS-DOS, one that they fully owned and controlled, not licensed. An IBM owned operating system would cut Bill Gates and Microsoft out of the picture entirely. If IBM could successfully pull off this two-fold attack, they would ensure that future PCs would be built off of hardware they licensed and controlled, and would run an operating system that IBM owned.
Any would-be cloner would have to come to IBM, hat in hand, and license both the operating system and the hardware it ran on from them, paying substantial royalties for both. This would ensure a steady stream of revenue flowing into IBMs coffers, something that was increasingly needed as the mainframe business they had built the company off of was beginning to look...a little shaky. Not that IBM had financial troubles yet, but it didn’t take an accounting genius to see that it would be difficult to sell mainframe computers for 100 thousand dollars if their job could be done just as well by a three thousand dollar PC. Especially if that PC wasn’t even sold by IBM.
Additionally, if IBM felt like it, they could simply refuse to license the hardware and software at all, ensuring that only IBM computers were sold and completely destroying the IBM compatible clone market. Steve Jobs did something similar to this ten years later when he returned to Apple, destroying the Macintosh clone market entirely and ensuring Apple’s complete control of its hardware and the software that ran on it. He did this by refusing to issue new licenses to cloners for Apple's new operating system, Mac OS 8, forcing existing ones to keep using the previous Mac OS 7, ceasing licensing hardware, and buying back the clone license for the largest Mac cloner, Power Computing.
In the words of a 1997 Wired article: “The other clonemakers...will be cut loose, allowed to continue only with products for which they currently hold licenses. They will not be licensed to sell new Apple technology. "We have no plans to give them Mac OS8 ... or the new Apple-licensed proprietary designs". The end result of this was that Macintosh
clone market ceased to exist in its entirety within a very short span of time. With the exception of a few rogue startups like Psystar, that are usually sued out of existence the moment they pop up on Apple’s radar, there are no companies making Macintosh computers apart from Apple. If you want to buy a Macintosh computer running the latest version of OSX, there is one company to get it from. So IBM’s strategy was not a bad one...in theory. IBM’s solution to its first problem, replacing the existing PC standard bus with a proprietary one, was also a clever idea in theory. Since the existing bus architecture was an open standard, there would seem to be no way of collecting any royalties whatsoever, nor of regaining control of it. But IBM had a *insert Baldrick clip of cunning plan* to fix this problem. They would
create a brand new standard, called Micro-Channel, that would be fully patented and owned by IBM. The incentive to switch to it would be the fact that it was far faster than the old ISA bus standard. This plan was the brainchild of Bill Lowe, one of the main figures responsible for the original IBM PC who we met in Part 1 of this series. The flagship for this shiny hardware
standard would be an entirely new line of computers, split into four different models. The first one would be an upgraded version of the original IBM PC 5150, using the full 8086 processor instead of the somewhat crippled 8088 variant that the original PC had used. Apart from that it would be fairly similar hardware wise, and would still use the existing ISA bus, maintaining compatibility with existing hardware add-ons. The next two models would be based around the 286 processor, a processor that Microsoft chairman Bill Gates had called “brain dead”, and the final one would be IBM’s first 386 based personal computer. With the exception of the first
model that was really just the original 5150 all over again and used the old bus, the other three would all solely be available with the new Micro-Channel standard. All four models would also come with the new 3.5 inch floppy drive as well, capable of holding far more data than the older 5 ¼ inch floppy drives found in the original PC and its AT and XT successors. IBM also had replaced the BIOS chip that Phoenix Technologies and Compaq had worked so hard to clone, with a new BIOS called the Advanced BIOS or ABIOS that would, in theory, somehow be invulnerable to reverse-engineering. These new systems would be called the Personal System/2s, or PS/2 for short, and IBM spent years designing them and creating and implementing the new Micro-Channel specifications, as well as doing joint development work with Microsoft as its junior partner on the second piece of its strategy, the new operating system that would replace MS-DOS on the PS/2 line, the synergistically named OS/2. More on OS/2 in a minute. In April of 1987, IBM publicly announced and showed off the PS/2 family. In Blue Magic: The People, Power and Politics Behind the IBM
Personal Computer, a contemporary book released in 1988, the authors give a good insight into how the new systems were perceived at the time, saying “For a change, the products were truly impressive” but admitting several paragraphs down that while the PS/2 line could “use most software available for IBM-compatible machines...it is not an industry-standard computer. Its Micro Channel bus was not compatible with add-in devices like the circuit cards available for IBM-standard computers.” This was a major shift in the PC market, and a very risky one. IBM was asking the entirety of the PC marketplace to buy machines that, for all their power, were completely incompatible with any add-in cards that the user had already purchased over the previous six years. And this was no small ask.
Today computers typically have only one or two add-on cards installed, typically a video card. Even further, the vast majority of computers sold today have no add-on cards with video, audio, and input/output options all integrated directly into the motherboard. In the area of laptops or tablets, it isn’t really even an option to do further upgrades in many cases. But the 1980s was a very different time for personal computers, and motherboards of that era frequently needed extensive plug-in cards in order to do things that today we either have on the motherboard or don’t even need. A few examples of fairly common PC add-on cards would be things like hard drive controller cards, audio cards, Ethernet or token ring cards, modem cards, MIDI cards, additional RAM cards, real time clock cards, serial, joystick, or parallel port cards, and many others. There were actually so many add on cards needed for IBM PCs that some companies,
like AST, got their start by designing and selling multi-cards, a single card that integrated multiple expansion cards into one drop in card. For example, AST’s popular SixPak Plus was an expansion card that integrated six different functions onto a single card, an RS-232 port, a parallel printer port, a battery backed calendar and clock, a game port, and an additional 384 kilobytes of RAM in order to max out the original IBM PC to its full 640k of RAM. This early multi-function card catapulted AST to 13 million dollars in sales in 1983, and gave them the platform needed to expand into other areas of personal computing, eventually launching their own line of PC compatibles and entering the Fortune 500 list in 1992 at spot 431. As a side note, the 8 Bit Guy has a great video on working tech support for AST that I would highly recommend watching as AST was a significant player in the 1980s and early 90s personal computer market, before they rapidly went downhill and went defunct entirely in 2001. With the PS/2 line and its Micro Channel bus, IBM was asking people to throw away years of investment and potentially thousands of dollars, in many cases exceeding the original cost of the actual computer itself. A person might have bought an original IBM PC 5150 in 1981, spent a bunch of money on add-on cards and then upgraded to a Compaq Deskpro 386 in 1986, carrying forward all of his expansion cards and adding some new ones.
Additionally, to add insult to injury, the new Micro Channel cards were not only incompatible with the old standard, they also cost more. A user would not only have to buy all new cards to get most of the same features they had in their non PS/2 machine, but they would pay even more per card for the privilege. If it had been anybody other than IBM trying to push this new incompatible standard into the marketplace, it would have been, in the words of a contemporary Byte magazine article, “hailed as innovative, clever, technically progressive, and destined to fail due to lack of IBM compatibility.” But this was IBM,
the creator of the PC standard in the first place, and the largest and most influential computer company in the world. If they could successfully move everybody over to the PS/2 line, from a hardware perspective there would be no more clones cutting into their market share. The whole PS/2 architecture and especially the Micro-Channel bus were all fully locked down and proprietary. Any company who tried to copy it without paying IBM’s steep licensing fees would be sued into oblivion, and the licensing fees would be high enough that it would be very difficult if not downright impossible for a would-be cloner to undercut IBM on price.
While IBM’s standard licensing agreement in the past had been a mere one percent of the final product’s selling price, the new licensing agreement would require a royalty of up to five percent of the final selling price, adding about 150 dollars or so to every PS/2 compatible. Furthermore, IBM also had what it considered to be a clever trick up its sleeve. To date it had been unable to match the frantic development rate of the cloners, who brought out entirely new generations of computers every 12-18 months. Unable to match this speed thanks to corporate inertia, IBM thus kept getting beat to market by cloners such as Compaq, who beat IBM to market with a 386 based system by almost seven months. Sure the new Micro Channel bus was locked down, but IBM decided to give itself a leg up and get twice the computer generations for the same effort. And so, the first generation of PS/2 computers would ship with the Micro Channel bus deliberately hobbled to only half speed. This would give IBM an easy way to,
in the words of Robert Cringely’s Accidental Empires, “discover all sorts of neat additional Micro Channel horsepower, which would be presented in a whole new family of machines using what would then be called Micro Channel 2.” They could produce a second generation PS/2 line that was twice as fast as the original, with essentially zero effort, allowing IBM to compensate for far lengthier development processes. In IBM’s mind, there was no way this strategy could possibly backfire on them. They were confident that the consumer market would follow them to the brand new standard, cutting the legs out from under every cloner and putting them in a position to either go out of business or come to IBM, cap in hand, and beg for a license to use the Micro Channel architecture. A license which may or may not be forthcoming.
The 1987 PS/2 reveal was accompanied by the reveal of OS/2, IBM’s next generation operating system that would be replacing Microsoft’s DOS operating system and the second part of their plan to regain control of the PC market. But IBM picked a funny way to create MS-DOS’s replacement, as they partnered with none other than Microsoft to develop it. The reasons for this strange partnership are somewhat unclear. One contemporary rumor was that
Bill Gates had heard that IBM was going to develop a replacement for MS-DOS alone, and had threatened to pull IBM’s license for MS-DOS if Microsoft was not included in the development process. Another possibility is that IBM was wary of going at it alone in the first place, with their confidence somewhat shaken by their failed TopView system, an MS-DOS replacement developed solely in-house that they had brought to market in 1985, where it had promptly crashed and burned. TopView is itself worthy of a full video, but for the sake of time we will just hit its highlights. It was a shell that ran on top of MSDOS, allowing for limited cut and paste between programs, as well as giving MSDOS a basic form of multitasking. In theory, the user could
open WordPerfect and work on a document while waiting for Lotus 1-2-3 to finish recalculating a spreadsheet. In practice however, TopView required such a massive amount of resources that multitasking really only worked on very light applications, and even then it was very sluggish. Additionally, since TopView was solely character based, it was unable to do a true windowing based GUI, since things like mixing graphics and text together or displaying varying fonts on screen inside of a document were not possible with a solely character based system. A quick note: although some of my sources state that TopView was character only, the Wikipedia entry for TopView says that it did have a graphics mode, it just was almost never used. So far as the user was concerned, they were still using MSDOS, only it ran considerably slower than standard MSDOS, required a more powerful computer, and didn’t really add much to their user experience. TopView also certainly didn’t resemble the new hotness of Apple’s Macintosh in any way whatsoever. Windows, although in an embryonic form in 1985, with version 1.0 coming
to market in November, at least sort of looked like a Macintosh if you squinted hard enough, and required 99 dollars to buy. Even still, interest was very low and Microsoft sold very few copies of Windows 1.0. What few people were interested in a more advanced MSDOS bought Windows and TopView was left to languish in obscurity. A few companies, such as MicroPro, invested a lot of time and resources into developing programs that used TopView’s abilities in much better ways than standard MSDOS programs, and they got badly burned with its failure as well. The press did not give TopView any good publicity either, with reviewers such as John Dvorak labeling it “TopDog” and not in a complimentary fashion. In Merrill Chapman’s entertaining book In Search of Stupidity, he recounts an entertaining personal story from when he worked for MicroPro and had to demonstrate a TopView version of their word processor, Easy, at a conference. A bunch of Macophiles from the press had for
some reason shown up and taken a great deal of pleasure in torturing me during demos of Easy. “Show us again how you cut and paste text,” they’d say. “You call that easy?! Bwahahahahahahahaha!” They brayed as I banged on a keyboard instead of brandishing a mouse. “Now how do you display a font onscreen? You can’t?! Wow! We can see why TopView is so fabulous!” they howled as they held their sides and laughed hysterically. By the end of 1985, it was clear that TopView was a complete failure. As Chapman states when recounting a conversation he had with upper management at MicroPro, sometime after his disastrous Easy demonstration: “Customers don’t like it. They’ve seen the Mac; that’s what they want on their PCs.
The press hates TopView; they think the Mac is where it’s at. Developers hate TopView; they want to make cool Mac-like things for PCs. Everyone hates TopView!” Faced with such a monumental failure, IBM’s confidence was probably at least slightly shaken. No matter how much IBM loathed their dependency on Microsoft, most likely they decided that they would be setting themselves up for failure if they tried to develop a DOS replacement all on their own again. A better plan in IBM’s mind, was to utilize Microsoft's expertise to
co-develop a successor MSDOS, but ensure that it was one that IBM owned in its entirety. Whatever the reasons for their decision may be, the fact was that on August 22, 1985, IBM and Microsoft signed a “Joint Development Agreement” that stated that the existing MS-DOS operating system would continue to be supported while its replacement was jointly developed, a replacement that was called “CP-DOS” in the JDA, but would eventually be named OS/2. IBM may have been developing OS/2 jointly with Microsoft, but they made it very clear that OS/2 was their product and Microsoft was at best a junior partner. IBM would retain full ownership
of OS/2, with the right to license it out if they chose or freeze out the cloners entirely. OS/2 was going to boot MSDOS out of the game entirely, relegating it to obscurity, hopefully along with its annoying founder and company. In the words from a 1988 Byte article called “OS/2 Dreams”, it would be “either the operating system that will take us into the twenty-first century, or the operating system that will take Microsoft into Chapter 11.” Byte missed the fact that as far as IBM was concerned, they wanted OS/2 to do both, and saw no reason why it wouldn’t. If Microsoft couldn’t license out MS-DOS anymore as everyone
switched to OS/2, then their biggest cash cow would be dead, and they would be relegated to competing with application software or games, just like any other software company. Fed up with seeing a steady stream of royalties roll into Microsoft’s coffers from licensing out MS-DOS to the very cloners that were killing IBM’s market share, they were determined to not only put an end to this torrent of cash, but also to make sure that OS/2 corrected all of MSDOS’s manifold weaknesses and was designed to the highest IBM standard of software development. Since IBM gauged its programmers productivity by the KLOC, or per thousand lines of code, OS/2 was certainly going to have an enormous amount of custom code and would be as far away from a lean, mean operating system machine as possible. The full development and history of OS/2 warrants a completely separate video of its own, so we will just focus on its early history here.
OS/2 was planned to be a multithreaded, multitasking operating system that could handle up to a whopping 16 MB of RAM, a number far in excess of MSDOS’s puny 640k. And yes it's true that by the mid 1980s you could have more than 640k of RAM in your PC compatible, but then you had to struggle with software and compatibility headaches as you used a clunky workaround like expanded memory, which clunkily allowed for access to extended memory above the 640k by means of bank switching RAM. A great read on the details of the 640k problem and the workarounds developed is over at filfre, check it out from the link in the description below.
IBM’s lofty ambitions for OS/2 immediately clashed with the reality of two companies with wildly different cultures struggling to mesh together to jointly develop a massive piece of software. When it came to the original MS-DOS, IBM had simply licensed it from Microsoft, they did no development on it themselves. But for OS/2, with IBM and Microsoft both writing the code, trying to coordinate who was working on what, and keeping everybody focused and working in the right direction was an almost impossible task. When development of OS/2 began in 1986, immediate cracks in the partnership’s facade began to show almost immediately. The corporate
culture clash continually exacerbated enormous tensions that frustrated everybody involved. In the words of Neal Friedman, one of Microsoft’s programmers: The project was extremely frustrating for people at Microsoft and for people at IBM too. It was a clash of two companies at opposite ends of the spectrum. At IBM, things got done very formally. Nobody did anything on their own. You went high enough to find somebody who could make a decision. You couldn’t change things without getting approval from the joint design-review committee. It took weeks even to fix a tiny little bug, to get approval for anything. Development crawled along at a glacial pace,
and matters were not helped along by the fact that although IBM and Microsoft were partners, the final word on all design decisions and differences of opinion always came from IBM as Microsoft was clearly viewed as the junior partner in the relationship. And IBM was constitutionally incapable of making rapid decisions, or reaching any sort of consensus in a timely manner. Rather, their byzantine policies and procedures and plethora of sign-offs from multiple levels of the bureaucratic hierarchy, ensured that any decision consumed enormous amounts of time. And when building an operating system, there are many, many, many decisions to make.
And every one ate up an inordinate amount of time. Matters were not helped by IBM’s decision to incorporate what was known as SAA or “Systems Application Architecture” into OS/2. The idea behind SAA was that OS/2 would be somewhat of a cross platform operating system, capable of running on the full range of IBM products, from personal computers all the way up to mainframes with either no or only minimal changes needed. This would allow programs written for OS/2 to run on any IBM machine large or small, giving IBM a further lock on the market from mainframes to micros. There was some precedence for this thinking, after all IBM had seen enormous success with the 360 line of computers back in the 1960s, a full range of computers, although obviously not including PCs, that all ran the same software and were compatible with each other, making moving software from one 360 system to another very straightforward. But times had changed dramatically in the past couple decades,
and the market was moving away from expensive high end mainframes with expensive support agreements and heavily marked up hardware, towards fleets of cheaper PCs linked together with network cards and a shared server. Trying to shoehorn SAA into OS/2 would not only dramatically increase the development time and workload, but would be somewhat similar to designing a new car that was also capable of being hooked up to a team of draft horses, just in case you wanted it to be compatible with older forms of transportation. Still, IBM insisted on this and thus work on OS/2 dragged on for years at an ever slower pace while the code base for OS/2 expanded at an exponential rate thanks to feature creep and IBM’s own development processes.
In 1987 OS/2 finally limped onto the market. And what had all of the development work produced after years of effort? A massive resource hog that, although under the surface contained many improvements over MSDOS, presented the user with the same cryptic command prompt that MSDOS did. All of OS/2’s fancy new features and design were not immediately evident as the project had gotten so far behind schedule that the actual graphical part of OS/2 was not ready to ship with the rest of it, and thus the initial release was command prompt only. But even though it was command prompt only, without the overhead of supporting a full bitmap graphical user interface, OS/2 was extremely slow. And not just slow, just to run it at all required a massive increase in RAM, hundreds of dollars worth of it. For an additional 340 dollars, more than three times the cost of MSDOS, which most likely came bundled with your computer anyway, you got an operating system that according to Byte’s tests, required 5 to 8 megabytes of RAM to run comfortably. This, in an era when most personal
computers rarely had more than 1 to 2 megabytes of RAM, and many Apple IIs and Commodore 64s were still being sold with only a fraction of that. The pricing also ran contrary to the expectations IBM itself had set six years earlier, when it released MSDOS 1.0 for a retail price of forty dollars. Another source, Merrill Chapman’s In Search of Stupidity says that OS/2 needed “4MB of memory to be useful, 8MB to step along smartly, and 16MB to really hum.” And as a reminder, if you really wanted a GUI for your compatible, you could always buy Windows for 100 dollars, version 2.0 of which was released in December of 1987. And Windows 2.0 only required a minimum of 512k of RAM to run, although as usual, anybody who tried to run it with the minimum specs quickly wished they weren’t. The Macintosh itself had been released with only
128k of RAM, which was barely sufficient but at least its GUI operated in it, and it's quickly released successor, the Fat Mac, had a total of 512k which allowed the Mac OS to run extremely well. Of course the Mac OS was not a multitasking OS at that time, and it was only black and white, however it was a fast and easy to use operating system so long as you respected its limits. IBM wanted users to pay 340 dollars for an operating system that potentially required 1600 to 2800 dollars of further investment in additional RAM. That was a lot of money, equivalent to 3400 to 6000 dollars in 2020. If Microsoft had decided to release Windows 10 with the requirement that it needed at least 128 GB of RAM to run comfortably and really should have at least 256 GB, it probably would have struggled in the marketplace. Although one tangible benefit of running Windows with that much RAM would be the ability to open at least four concurrent tabs in Chrome, something that many people would doubtless appreciate.
IBM also did nothing to clear up the confusion many people had with the OS/2 and PS/2 naming scheme, where people assumed that OS/2 would only run on the PS/2 line of computers. Of course IBM had no interest in clearing up that confusion when the double punch of OS/2 and PS/2 meant that people would of course only be buying PS/2 personal computers anyhow. A number of companies did hop onto the OS/2 bandwagon however, primarily smaller companies that were hoping to launch into a less crowded marketplace than the MSDOS world, or companies that were beginning to feel that dealing with IBM was preferable to dealing with Microsoft. There were a few heavy hitters in the latter group, such as Ashton-Tate and Lotus, who fully committed themselves to OS/2 development, and ignored Windows. This would wind up painfully biting them in the long run. The PS/2 line was launched into the marketplace with a muddled ad campaign that was the exact opposite of their extremely well received ads for the original PC. Whereas the ads that
introduced the original PC to the world featured Charlie Chaplin’s famous Tramp character and were extremely clear and easy to understand, the PS/2 line would be introduced by...a number of cast members from the famous TV show MASH. Although they did not appear in character, Harry Morgan, Loretta Swit, Gary Burghoff, Larry Linville, William Christopher, Wayne Rogers, and Jamie Farr appeared in a series of TV commercials and print ads extolling the wonders of the PS/2 line. Although developed by the same marketing agency that had developed the original Little Tramp ads, Lord, Geller, Federico, Einstein, the contrast between the work they had done years prior with the original IBM PC commercials and the new PS/2 commercials was striking. The Tramp commercials had worked well due to the fact that they featured a single extremely recognizable character played by Charlie Chaplin, they did not feature the actor himself.
In an AP article on the new ads, the following odd justification is given: Richard Lord, chairman of the ad agency, said the company and the agency were uncertain whether the Chaplin character ″could carry all the products″ being introduced in the new line of computers. He said the idea of computer ″connectivity,″ or the ability of computers to work together, had also become a more important consideration than it had been six years ago. Using a group of actors in the ads, Lord said, would give the admakers the opportunity to show computers being used in a variety of situations. Having most of the primary cast of MASH do advertising, but not as their famous characters, was a very odd marketing choice, and one that left many people unmoved. At least it wasn’t actively insulting the audience, as Apple’s infamous “Lemmings” commercial had done several years prior. Ultimately, IBM felt that they had been so poorly served by their advertising agencies, that they eventually terminated their relationship with all of them, shutting down over 500 million dollars a year in ad accounts. But that was well after the damage had already been done.
In fairness, bad marketing was undoubtedly the least of IBM’s many problems with the PS/2 line. Even good marketing would have struggled to deal with driving consumer demand for a product that was not only expensive, but also was completely incompatible with all of their expensive hardware add-ons. For some reason, IBM failed to think about the problems that this would incur, especially when the first generation Micro Channel bus was actually slower in many cases than many clones using the old ISA bus. Trying to make consumers buy a new computer that required them to give up their existing investment in add-on cards and buy all new replacement add-on cards that were not only more expensive than the old ones, but were actually slower as well seems more like a Dilbert cartoon than an actual strategy but that was what IBM decided to do. In fairness, the PS/2 line did require somewhat fewer add-on cards than its predecessors, as PS/2 motherboards had a decent amount of functionality built in that previously required separate add-ons. Most significant
was putting video out directly onto the board, with the new Video Graphics Array port, or VGA that was built into all PS/2 motherboards. However most users who had expanded their ATs would still have needed to, for example, replace their ISA sound cards with new MCA ones. And some users would have just been out of luck entirely, as some of the more esoteric ISA add-on cards did not exist in Micro-channel variants, and never would.
Furthermore, IBM decided that not only would PS/2 cloners have to pay license fees of 5 percent on every Micro Channel clone they sold, they would also be required to pay a retroactive 5 percent license fee for every single PC clone they had previously built using the ISA bus. Cloners would have to cough up large sums of money to not only use the new architecture, but would also have to retroactively pay IBM millions for computers they had already built and sold using the open PC bus that IBM did not own the rights to. At this point, even Dilbert’s Pointy Haired Boss might be saying that this sounded like a bad idea. Bill Lowe, head of IBM’s Entry Systems Division, started 1987 by confidently waiting for cloners to approach him about licensing. Publicly, he stated
that any cloner who tried to reverse engineer Micro Channel would need to either have great engineers or very good lawyers. He was confident that it would be neither feasible nor legal to clone it, and thus the would-be cloner would be at his door trying to get a license, which IBM would probably be unwilling to give, at least initially. As the months went by and no cloner made any overtures, Lowe started publicly saying that IBM might be willing to license the design out to cloners for the right terms. Still nobody came. By 1988, Lowe was, in the words of one author “practically begging other companies to adopt the Micro Channel. How could IBM claim
to own the new standard if nobody else used it?” It wasn’t that the cloners were ignoring the PS/2 line and Micro Channel, Compaq even quietly did a decent amount of work towards engineering a clone, but without wide market adoption, there was little incentive to quit producing machines using the old ISA bus. And those machines were what was selling, with the broader consumer market treating the PS/2 line and its complete incompatibility with their existing hardware with complete indifference. Still, if Micro Channel had been running at its full speed, it would have been at least twice as fast as the older bus, and that might have been enough to overcome IBM’s poor decision making by making it attractive enough for consumers and businesses in broad numbers to stand for the extra expense and buy the new machines and rebuy all their hardware expansion. But hobbled as it was, plus the massive fees to use it and the forcible extraction of royalties that IBM had no legal right to...the existing cloners were up in arms and they had the powerful force of
the broader consumer market behind them. The consumer market wanted “PCs based on the powerful new Intel 80386 chip or they wanted good old reliable machines along the lines of IBM’s AT.” And IBM was offering neither option in a format the market wanted as the PS/2 386 line was incompatible with consumers existing hardware and the popular AT had been killed off by Bill Lowe. Yes, IBM actually killed off their top selling computer, the AT, in an attempt to force
market adoption of its incompatible replacement. Shockingly this bold strategy didn’t work out, especially because the market could still buy plenty of AT clones from Compaq, Dell, or any of the plethora of other cloners. Fortunately for IBM, Lowe’s cancelation of the AT in North America did not affect IBM’s operations in Europe, where Bill Lowe’s previous boss, Mike Armstrong, had recently moved to take over IBM’s European branch. Armstrong wisely
kept the AT in production, and it continued to be highly successful in Europe for quite some time. IBM was correct in diagnosing the problem with the ISA bus, with new faster processors like the 386 and its 32 bit architecture, a new bus was definitely needed in order for that computers equipped with the 386 and future 32 bit processors would not find themselves hobbled by a 16 bit bus. IBM had failed to make Micro-Channel that new standard, but there was another company with a replacement bus that was ready to take its own stab at the problem. And its success would see IBM’s chances of regaining control of the PC standard permanently slip away from its grasp. Unsurprisingly, the company that would lead the rebellion against Micro Channel was none other than Compaq. Compaq had been quietly working on a replacement bus that was not only
as fast as Micro Channel, but was also fully backwards compatible with the legacy ISA bus, the very thing that IBM had said was impossible. Using this proof of concept, Compaq reached out eight other significant cloners, creating an uneasy alliance that the Wall Street Journal called The Revolt of the Clones. This alliance’s goal was to take on IBM and force the adoption of a Micro Channel alternative. This consortium was comprised of Compaq, AST,
Epson, HP, NEC, Olivetti, Tandy, Wyse Technology, and Zenith Data Systems and had a combined 1987 market share of 33.5 percent of the PC market. Although led by Compaq, this was definitely a fractious alliance of fierce competitors, brought together solely due to the need to present a united front against IBM and avoid being put out of business by IBM’s new standard. Out of this unusual alliance of bitter competitors emerged a new bus standard, the Extended Industry Standard Architecture, a full 32 bit bus that did what IBM claimed was impossible...it was fully backwards compatible with the old AT ISA bus. IBM had now definitely lost control of the PC hardware standard and they would never regain it. IBM’s attempt to break the cloners had backfired badly, as one source bluntly put it: “(Bill) Lowe’s PC strategy had been a torpedo below the waterline at IBM. No one knew it by the end of 1988, but IBM was sinking.”
IBM had lost its battle to force a new industry hardware standard and regain control of its hardware. However, if it could succeed with OS/2, it could at least regain control of the operating system that every PC ran. Given the rapidly decreasing margins on hardware, controlling the software the hardware ran on would be a great consolation prize, and would allow IBM to replace Microsoft. Software had extremely high margins, and OS/2’s success would firmly relegate
Microsoft to merely being an application company with its productivity applications such as Word and Excel, probably a large and influential one, but not the dominant one. Although interest was initially high for OS/2, its repeated delays prior to its first command line only release, plus even more delays in the release of its graphical user interface, plus the massive cost of the memory required to run it, meant that people kept right on using MS-DOS, and many started to pay more attention to Windows. And IBM further shot itself in the foot with its stubborn refusal to leverage its vast RAM production capacity to help further the installed base of OS/2 and get it to a critical mass. Given the massively prohibitive cost of the RAM required to run OS/2, IBM could have started bundling OS/2 for free or massively discounted with RAM sticks. In the words of a product manager for an OS/2 word processor called DeScribe, “OS/2 without memory was a one thousand dollar upgrade. Bundled with a handsomely discounted 4MB memory stick, it was a million copy seller” Because the price of RAM had quadrupled in the late 80s, with 1 MB sticks going from 100 dollars to 400 dollars, IBM would have profited handsomely from giving OS/2 away with RAM and driven market adoption of OS/2. Sadly for IBM, nobody at the
top ever thought about adopting this strategy. On May 27th, 1988 Microsoft even released a special version of Windows 2 that took advantage of the powerful new 386 processor to allow for multi-tasking multiple DOS programs, if you had the RAM for it. It also gave the user many of the features of OS/2, although it was still behind the Mac, Amiga, or even the Atari ST in many areas. Although still three years away from the 3.0 release that would rocket it to the moon, Windows 2 gave the user a GUI that ran well in a single megabyte of memory and only required 512 kilobytes as minimum specs. For users who wanted a GUI for their DOS machines,
they could pick OS/2 which cost a fortune and required a larger fortune in RAM to run well. Or they could go with Windows, which was reasonably priced, pirated enough that many got it for free, and ran acceptibly, especially if you had a blazing fast 386 machine. Microsoft didn’t really care too much about Windows being pirated back then, the more people using it the more market share it got, and Microsoft made most of its Windows money licensing it out to OEMs for preinstall on new systems anyhow. OS/2 development was also problematic, as the pricing for the software development kit or SDK had been set at 3000 dollars, a price greater than even Apple’s SDK’s and one that stood in stark contrast to the Window SDKs, which Microsoft practically gave away. For large companies this was not a huge problem, but for small
companies or individual programmers, the initial investment required was a substantial barrier. IBM also failed to set up any programs to help promote OS/2 products, making no attempt to secure shelf space on retailer’s shelves or help software vendors market directly to OS/2 users. One other problem with OS/2 is worth mentioning. When IBM finally released OS/2 1.1 in October of 1988, with the Presentation Manager included, finally giving OS/2 its graphical user interface and making it look far more polished...it couldn’t print. More specifically
it couldn’t print to any non-IBM printer. IBM had made no effort to get drivers for any other printers besides the ones it sold, and most people who had printers didn’t have an IBM one. Printers in the 1980s were far more expensive than today, and represented a significant investment for the user, especially if they had one of the new laser printers. This meant that a new user
of OS/2 was not only confronted with needing to replace all of their hardware, buy OS/2 for 340 dollars, pay a small fortune for enough RAM to make it usable, and deal with a pronounced dearth of OS/2 specific software...but they also needed to buy a brand new printer from IBM. Windows was looking more and more attractive by the day, Windows 2.0 started selling 50,000 copies a month. And Windows 2.0 could actually print to non-IBM printers of course. In October of 1989, IBM released OS/2 1.2 which still struggled to print to non-IBM printers. Sales of Windows 2.0 rose even further thanks to this, and then in 1990, Microsoft released Windows 3.0 which promptly sold two million units in six months and pretty much sealed
OS/2 into a permanent vault of obscurity. Although the PS/2 line of computers was a market failure, there are a couple hardware holdovers from it that continue to the present day in some form. First of all was the PS/2 serial port for a mouse or keyboard, this small analog port, typically purple for keyboard or green for mouse connection is still found on some towers even today. Second was the 15 pin VGA monitor connector, which became a standard
for connecting a monitor to a laptop or tower, and is also still found on some towers today, although it has essentially been entirely superseded by HDMI or the somewhat older DVI. OS/2 actually soldiered on for well over a decade after its initial release, its final release was in December of 2001, with IBM’s support of existing installs continuing until December 31, 2006. Never quite managing to break through IBM’s incredible ability to shoot itself in the foot, it still attracted a dedicated user base that was legendarily devoted to it and its stability and reliability. Scott Adams even refers to OS/2 users in one of his Dilbert comic strips from 1995, poking fun at the fact that OS/2 die-hards were even more fanatical about their operating system than Macintosh users were. That wraps up part three of the rise and fall of the IBM PC. I had a lot of fun researching each part and writing the scripts, and I hope you enjoyed it. The next computer history series I will be starting sometime in the next year will be
an in-depth look at the rise and fall of OS/2, possibly interwoven with the rise of Windows. Both subjects have been touched on in this series, but both deserve their own stand alone videos. If you want to see more videos along these lines, drop a comment below with what computer history subject you would most be interested in seeing. And as always, thanks for watching! If you enjoyed this video, please leave a like and consider subscribing.