The following article is the script for the Fall of OS/2 video, which you can watch right now on the Another Boring Topic YouTube channel. I do want to make it available as an ePub/Mobi file but I am still figuring out how best to do that.
Today we will be finishing the story of OS/2, starting from the 1990 release of the final OS/2 version that Microsoft and IBM developed under the 1985 Joint Development Agreement. Following this, Microsoft and IBM acrimoniously split, and IBM took over sole development of OS/2. IBM’s then continued attempts to crush its erstwhile partner and drive Windows from the marketplace, or at least make OS/2 a viable competing standard to Windows was, in the end, an effort doomed to failure.
This wasn’t due to any superiority that Windows had over OS/2 in features, but rather IBM’s stubborn refusal to do anything intelligent to get OS/2 into the hands of consumers, and indeed IBM’s almost Commodore-like ability to shoot itself painfully in the foot as frequently as possible. I say almost Commodore-like because IBM is still around today and Commodore…well I’m sure they are having fun romping around a meadow with IMSAI, MITS, and maybe Digital Research.
Whereas Microsoft did almost everything right when it came to getting widespread Windows adoption, IBM did almost everything wrong when it came to trying to do the same for OS/2. But before the end came, OS/2 somehow soldiered on, getting progressively more mature and feature rich, and inspiring a following that was arguably even more devoted than Apple’s acolytes, as Scott Adams referenced in a 1995 Dilbert comic strip.
It has been said that while success has many fathers, failure has only one. Yet OS/2’s failure required a host of fathers, apparently assiduously working towards a single goal…preventing OS/2 from ever truly competing with Windows. One might almost suspect that Microsoft had infiltrated a legion of nerdy James Bonds into IBM’s upper management, with a mission of ensuring the complete domination of Windows. Sadly there were no secret agents involved, just a long succession of incompetence from both individuals as well as IBM’s increasing corporate dysfunction and absolutely mind-bogglingly complex bureaucracy.
To illustrate this last point, I will simply point out that in 1988, an IBM vice president, one of the top fifty positions in the entire company…was still a whopping SEVEN layers away from the chief executive, so far as IBM’s reporting structure went. That sort of corporate structure isn't exactly a recipe for a fast moving company, but it absolutely is a recipe for enormous amounts of corporate paralysis, infighting, and dysfunction.
There is something that I want to say though, before we really start diving into things. I am about to be pretty hard on IBM as a company, and I am about to be pretty hard on a couple of IBM’s leaders. However I want to be very, very clear that I do not think a single person or division at IBM deliberately and/or maliciously sabotaged OS/2’s chance of being successful in any way.
Of the multitude of poor decisions that eventually wound up destroying OS/2, some were immediately evident as a bad choice, some were very quickly evident as a bad choice, and some were a bad choice simply due to bad luck for IBM mixed with good luck for Microsoft, but none of these choices were made with the deliberate goal of destroying OS/2.
The fact of the matter is that IBM was an massive organization that was rife with infighting and red tape, and there were any number of good leaders in it who attempted to succeed in spite of the dysfunction, sometimes while also attempting to reduce the dysfunction and red tape.
That’s a hard battle to fight, and even the best of leaders can fail to correctly juggle those two burdens. A certain Dilbert series about battling business units is very, very applicable here, as IBM’s various divisions and power blocs squabbled and fought for their internal goals and successes, with IBM’s success as a whole being a frequent victim, and no victim was more injured, more often, than OS/2.
To help balance out the story, I am going to, as I can, single out a couple people who attempted to recover, repair, or at least mitigate the effects of the poor decision making. One was an IBM leader who brought a level of energy, hard work, and experience to his role that in a more fair world, would have been rewarded with more success. The other man was outside of IBM, but was passionately devoted to OS/2 and did everything he could to promote it while also calling out its weaknesses in an attempt to rectify them.
In a more fair world his passion would have been better rewarded as well. I hope that by doing this, I can balance out the criticism I am about to levy with at least a hint of positivity. IBM genuinely had a lot of creative, hard working, genuinely interesting people within the company…and I am doing my best to avoid making a joke about none of them apparently being in the marketing department. And the OS/2 community as a whole had a lot of passionate supporters in it, who took OS/2’s ups-and-downs very personally.
On an unrelated note: while I’m not aware of any sources that state this, I have to at least point out that IBM’s naming scheme for OS/2 was really, really odd and seemed to have little rhyme or reason. As an example, OS/2 1.3 was sold in competition with Windows 3.0, and OS/2 2.0 and OS/2 2.1 competed against Windows 3.1. If consumers know nothing about two software products, there is a tendency to pick the one with a higher, or at least less confusing, number…although again I have never seen a source cite this as one of OS/2’s problems. However I guarantee OS/2’s odd versioning didn’t help position it in the marketplace.
OS/2 1.3
In December 1990, OS/2 1.3 was released. This version was the final one to be developed under the Joint Development Contract between Microsoft and IBM. OS/2 1.3 was far more usable than previous OS/2 versions, as it not only had a fairly high chance of actually printing correctly on the printers most people owned, but it also had a reasonably nice suite of applications available from IBM’s Desktop Software division.
And here we see IBM’s incredible ability to shoot itself in the foot, strongly manifest itself. We have seen it before with how IBM lost control of the PC market hardware standard, now it’s time for this tendency to start attacking whatever control IBM still had over the software standard…and the Desktop Software Division was about to be the first major victim of the 1990s.
Based out of Milford, Connecticut, this division had been formed in 1988 when IBM had the rather surprisingly bright idea of forming a new division of people, brought in from outside the polished suits-and-ties world of Big Blue. The division’s goal was to focus on developing more applications for OS/2. Being from outside the mainframe dominated world of Big Blue, they could bring in fresh ideas and perspectives, and hopefully help give IBM the knowledge it needed to successfully compete in the application marketplace
The Desktop Software division threw itself into the work with enthusiasm, and by the time OS/2 1.3 was ready for release in 1990, their hard work had produced a suite of products that enabled a new OS/2 user to do word processing, give business presentations, and even do desktop publishing. And these products were all reasonably polished, solid experiences. For some arcane reason I have not been able to figure out, these applications were also going to be released for Windows, however the Windows versions were at least scheduled to ship well after the OS/2 versions, with the exclusivity hopefully giving OS/2 a boost out of the gate.
However the Desktop Software division had not solely been tunnel focused on developing OS/2 applications, they had also been paying close attention to the marketplace and the rising Windows tide. Realizing that OS/2 needed some time to build momentum, and also recognizing that the PC world was eagerly seeking a modern GUI, they were concerned that Windows would steamroll right over OS/2 before it had a chance to build market share and meet the market’s desires. With this in mind, they managed to get a meeting set up with the president of IBM himself, John Akers.
In the meeting, the Desktop Software division representatives begged Akers to take the rather unusual step of buying a GUI from a small software company called GeoWorks, whose flagship product carried the same name. Unlike the legendarily sluggish Windows, GeoWorks ran quite well even on the slow 8086 processor of the original IBM PC from 1981.
This wasn’t the only option on IBM’s table either, they could also choose to throw Gary Kildall a bone and use GEM running on top of DR-DOS, cutting Microsoft and MS-DOS out of the picture. DR-DOS was an excellent operating system that, for all intents and purposes, was fully compatible with the existing body of PC software, and GEM was an excellent GUI that ran on top of DOS, whether DR-DOS or MS-DOS. While GEM itself would eventually power the Atari ST, sadly a pairing with IBM would never happen of course.
One final option that according to Merrill Chapman’s book, was widely discussed inside of IBM, was to use the Presentation Manager interface and then apparently slather it on top of DOS. Now, I cannot seem to find any details about how this would have worked, so I am somewhat guessing here. But the only way this would seem to make sense would be if Presentation Manager was a program that ran on top of DOS, similar to how Windows functioned.
Presumably, the eventual goal would have been to quietly migrate people over to OS/2, which would have looked the same to the average user, but had the modern OS/2 foundation instead of the aging DOS one. It seems like this wouldn’t have been a bad idea, but I’m curious what you all think about it. Let me know in the comments if you think this idea would have worked out better than what IBM actually did.
Even as the 1980s came to a close, IBM still had the upper hand with Microsoft, and could even go so far as to threaten them with the termination of the Joint Development agreement, and just go their own way if they didn’t halt all Windows development. At this point, which was prior to the release of Windows 3.0 and the first significant success Microsoft would see with Windows, there was at least a good chance that Microsoft would have backed down, potentially allowing IBM to strangle Windows before it had a chance to firmly cement itself as the dominant GUI for PCs.
But it was not to be. When the Desktop Software division representatives met with Akers, they were treated politely but condescendingly. In Search of Stupidity quotes Akers as calling them “a group of good kids” before lecturing them on the realities of life as IBM saw it. Essentially, IBM controlled the PC market, controlled the standards, and presumably thanks to OS/2 and PS/2 it always would. Akers referred to Gates as “a nice boy” who presumably was too innocent and naïve to ever pose a threat to IBM. Akers also confidently asserted that IBM had a cunning plan to solve the problem of positioning OS/2 versus Windows.
This plan would soon be realized in a Pythonesque agreement with Microsoft in which Windows was supposed to be positioned solely for “low-end machines” whereas OS/2 was reserved for the ratified heights of “high-end” machines. Or in other words, Windows was for the computers most people owned, and OS/2 was reserved for the computers they wished they could afford. Meanwhile back at the meeting with Akers, the Desktop Software reps were given a pat on the head and a lollipop before being sent back to their cubicles.
The entire Desktop Software division was then disbanded in 1992, right before the release of OS/2 2.0. This ensured that future releases of OS/2 2.0 were robbed of the applications that they would otherwise have ensured were available, something that would almost immediately hit hard with the next major release, OS/2 2.0. And as Windows accelerated into an unbeatable titan, proving that the concerns raised had been 100 percent valid…it was probably at least mild consolation to the former Desktop Software members that Akers was booted out of IBM shortly after disbanding them.
Jim Cannavino
I promised that we would try to balance out the criticisms by spotlighting a couple of interesting people from IBM, so I think this is a good place to introduce Jim Cannavino, the new head of IBM’s PC division as of the end of 1988.
Cannavino, although comparatively young, was a longtime IBM veteran, and was considered a potential candidate for IBM chairman. A hardworking, hard driving man, Cannavino had previously headed IBM’s mainframe division, responsible for half of IBM’s total revenue and a whopping two-thirds of its profits. And he did it all by age 40. Cannavino’s story is an interesting one as he was far from the typical IBM employee, having talked his way into a job interview when he was only a teenager. And when I say that Cannavino was hard working, if anything that’s an understatement.
He married as soon as he graduated from high school, supporting his young and growing family by working three separate jobs at the same supermarket: produce manager, meat manager, and a third managerial position. The three salaries kept him working from 6 in the morning until 10 at night, but it did let him earn a reasonable living for his family. Just for extra fun, on top of all of this work he also started taking electronics classes at the DeVry Technical Institute, near Chicago.
Eventually, while still a teenager, Cannavino found his way to an IBM branch office in Oak Park, and managed to bluff his way into an unscheduled job interview with the branch manager, a Mr Thompson. Cannavino convinced Thompson to give him the aptitude test for IBM’s repair technicians, and aced it. But since he only had a high school degree, he didn’t really meet IBM’s standards for hiring technicians. But since it was the early 1960s and IBM was in the middle of a rapid expansion due to its five billion dollar gamble on the new S/360 line of mainframes. Thompson took a chance and hired Cannavino, who again, was still a teenager at the time.
As Thompson was considering his decision, Cannavino promised him that he would be Thompson’s best repair technician within six months of being hired, or else he would give back all of his paychecks. Thompson thought Cannavino was joking…but he eventually found out that Cannavino was dead serious. How did he find out? A few months later when IBM’s payroll division reached to him to find out why Cannavino was throwing off the books by never cashing his checks. Cannavino explained to the first incredulous, then amused Thompson that he had been keeping all of his payroll checks in a shoebox, just in case he had to return them. Since this obviously meant that he had no income from IBM, he had taken a job at a pizza parlor on the side to pay the bills.
From that point on, Cannavino’s rise through IBM had been meteoric, as he soon moved from repair work into mainframe programming, quickly gaining a reputation as one of IBM’s best bug fixers. Before too much longer, he was moved to Poughkeepsie, New York, the center for all of IBM’s mainframe programming at the time. He soon figured out a way to make a lot of existing applications run much faster on the new S/360 systems, and when nobody believed him, he basically stole a mainframe to prove it.
Okay he didn’t technically steal it, what he did was first find an IBM division that was scheduled to have a mainframe delivered to it that he felt wouldn't miss it for a few months. Then he sneaked into the mainframe shipping area and changed the shipping label on the division’s shiny new mainframe, having it sent to a vacant IBM lab instead. He then spent the next several months working with a couple other programmers to prove his theory, and was able to make improvements that led to speed increases of over double for many applications.
This and other successes, plus Cannavino’s incredible appetite for work, rocketed him up to first the head of IBM’s mainframes, and then to take over the PC division after Bill Lowe decided to leave for a quieter position at Xerox in 1988. And after spending six months working on fixing some of the most pressing process issues in the division, Cannavino turned his formidable attention to Microsoft and what exactly IBM was getting out of its business relationship and the JDA. And he was not happy with what he found. OS/2 wasn’t selling well at all, it had a lot of bug reports stacked up, and Cannavino was not happy with what he felt was the undisciplined way Microsoft wrote code.
Remember, Cannavino was not just a mainframe programmer, he was the product of a programming culture that emphasized methodical, well organized, well commented, and painstakingly documented code. This was umm..not exactly the way Microsoft worked. And the clash between Cannavino’s methodical mainframe programming background and expectations and the more free wheeling, maverick style that Microsoft's programmers tended to use…well it was bound to cause immediate friction.
Additionally, a lot of key parts of OS/2 had been written by Microsoft, not IBM, and according to Big Blues: The Unmaking of IBM, “if IBM ever tried to make a clean break, it would find that it had to start pretty much from scratch on parts of the code that were not only important, but were so complicated that it might take IBM programmers a couple of years to get up to speed, a frightfully long time in the PC business.”
Cannavino also had to deal with the fact that IBM’s accounting procedures treated software development in a way that basically handcuffed him to OS/2, no matter his concerns. Essentially, IBM only counted a quarter of the yearly cost for developing a piece of software, with the remaining three quarters spread over the following years. By 1989, IBM had been spending about 150 million dollars a year on OS/2 development…and had rolled forward the bulk of that over a number of years, totalling at least 225 million of deferred costs hanging over Cannavino’s head.
If he decided to back out of OS/2 entirely, which was unthinkable anyhow based on the amount of public promises IBM had made to the businesses that had already invested into OS/2…but if he decided to do it anyhow, along with all the other problems, almost a quarter billion dollars of deferred expenses would come crashing down onto his head. It was unthinkable.
And so Cannavino decided he had no choice but to continue with OS/2 and with Microsoft as a partner, at least until he could figure out what to do to break the deadlock. Throughout 1989 and 1990 Cannavino worked at this problem, while he also did his best to court some major software makers like Lotus, Software Publishing, and Metaphor to invest into applications for OS/2, while also promising them that he would bring Gates to heel and ensure that Windows would not take over the market. And he had a series of intense meetings with Gates, starting in summer of 1989, trying to hammer out a compromise that would solve his problems, or at least reduce them.
But it was all for nought, in spite of Cannavino’s best efforts to figure out a solution that would enable him to crush the threat Windows represented and position OS/2 for success. In the end Microsoft and IBM went their separate ways in 1990, and IBM had to take over full responsibility for OS/2’s continued development and marketing, for better or worse.
But this almost didn’t happen. An interesting tidbit comes from the book Big Blues: The Unmaking of IBM. According to this book, in early 1991, Cannavino was seriously considering just throwing in the towel on OS/2, admitting Gates and Windows had won, and saving the 125 million dollars a year that he was spending on an operating system that very few people seemed interested in. And the then chairman of IBM, John Akers, was apparently thinking the same thing.
In a meeting with Akers and other IBM senior management in early 1991, Akers basically asked why IBM should continue to invest in OS/2 when it seemed the market just wanted Windows. Cannavino responded by admitting that maybe IBM should just give up on OS/2, however he thought that there was a sixty percent chance that it was worth continuing.
According to Big Blues: “Cannavino said that he thought that IBM should continue with OS/2 partly because it could make real money in personal computer operating systems, even though the market was a relatively small $600 million a year and IBM, through OS/2, had almost no market share. Cannavino also raised the whole issue of setting standards. If Microsoft controlled the market, then it, not IBM, would decide when to build capabilities into the operating system that would let software developers build video, for example, into their applications.”
Cannavino clearly had a pretty good idea of what was at stake, and he succeeded in persuading Akers to continue OS/2 development. OS/2 would survive another day. And, in spite of his reservations, Cannavino was determined to do whatever he could to make it a success.
One additional rather entertaining nugget that I found lies in OS/2’s reported sales numbers, specifically its claim by early 1991 that around 600,000 copies of OS/2 had been sold since its 1987 launch. IBM arrived at this figure by taking the 300,000 copies of OS/2 that were actually sold, and then adding another 300,000 copies of OS/2 that were given away to a number of IBM’s largest accounts when they purchased sufficient amounts of IBM’s memory boards.
According to Big Blues: “As far as anyone could tell, those 300,000 copies of OS/2 had never been used.” The interesting part comes in when Bill Gates started publicly pointing out that only 300,000 copies had been sold, something he would know because Microsoft had only received royalties for 300,000 copies of OS/2. IBM quickly sent him a royalty check to cover the other 300,000 copies…provided Gates agreed to start publicly using the 600,000 sales figure instead. Which Gates did…while also telling multiple reporters exactly what had happened and then smilingly insisting that they use the 600,000 figure.
OS/2 2.0
In spite of the collapse of the JDA, IBM’s developers got to work fairly quickly on getting the next version of OS/2 ready to go all on their own, and after a mildly surprisingly short two years of development, OS/2 2.0 was released in April of 1992. Largely thanks to the previously mentioned disbandment of the Desktop Software division, it didn’t exactly launch with a host of killer apps. Or even one killer app. This failure was again, entirely on IBM’s shoulders, as OS/2 2.0 was the first release that IBM was solely responsible for marketing and selling and they were badly struggling to figure out a coherent strategy.
By this point it was clear that OS/2 was badly trailing Windows in sales, which was racking up large numbers thanks largely to the success of Windows 3.0 and had sold around 30 million copies across all versions by this point. This number does not include the large number of pirated copies either, as contemporary versions of Windows were almost ludicrously easy to pirate, something that undoubtedly helped drive greater adoption.
And to add insult to injury…guess what else was released in April of 1992? Yep, you guessed it, on April 6 1992 Microsoft released Windows 3.1. So OS/2 2.0 was launched directly into strong headwinds as the successor to the most popular Windows version to-date Godzilla stomped onto the scene the exact same month. Even before its release, Windows 3.1 was already clearly going to be a massive success, and I doubt it's a coincidence that it was released just in time to stomp all over OS/2 2.0’s launch. Gates always plays to win, and he always goes for the throat whenever possible.
From a features perspective, OS/2 2.0 was a genuinely powerful release, with functionality and stability that were years ahead of Windows 3.1. It was a largely a 32 bit operating system, capable of exploiting the full power of 486 and the new Pentium processors whereas Windows 3.1 was still trapped by its 16 bit legacy.
OS/2 2.0 had also embraced object oriented programming, which was extremely forward looking and also something that Steve Jobs’ NeXT computer company had been pushing for years with their advanced NextSTEP operating system. I mostly bring this up to demonstrate that IBM was genuinely innovating and trying to push the state of the art forward in bringing a modern, multitasking, 32 bit, object oriented operating system to the broader consumer market.
OS/2 2.0 also ditched the old Presentation Manager graphical interface that had been developed in partnership with Microsoft, and contained many elements in common with Windows, in favor of a new and original graphical interface called Workplace Shell. OS/2 was moving firmly away from its confusing Windows/Microsoft heritage and was becoming a solidly IBM product. And honestly, this was a very good thing and OS/2 2.0 really does a good job of showing that IBM had learned a thing or two over the previous years about how to develop good software. Unfortunately everything apart from OS/2 development was still pretty much a die roll as to whether it was going to be a smart move or not.
To help drive the broader OS/2 community, IBM had founded the IBM Independent Vendor League, or IVL. This organization was formed to help market OS/2 through books, magazines, certifications, and similar marketing strategies. In addition, there were also several forums set up, most prominently being Will Zachman’s Canopus forum, which was hosted on Compuserve.
IBM also made a smart decision in appointing David Barnes as OS/2’s lead evangelist, a position that according to his LinkedIn, he began in January of 1990. Barnes was probably the closest thing IBM had to Guy Kawasaki, Apple’s chief evangelist for the Macintosh back in the 1980s. Barnes was an enthusiastic speaker who was not only passionate about OS/2, but was personable and extremely good at communicating OS/2’s strengths and, well, evangelizing for the platform. He was also extremely good at taking complex features such as multi-threading and 16 versus 32 bits, breaking them down and explaining them in an easy-to-understand fashion.
You can get a good feel for how good a speaker he was from the following clip, taken from a 1993 event where he presented OS/2 2.1 versus a Microsoft rep who presented NT, and then a lively Q&A ensued. According to his LinkedIn account, in his role as lead evangelist, Barnes traveled to 41 different countries, gave up to a dozen presentations a week and even starred in a series of commercials for IBM, called The Warped World of David Barnes.
He was a huge asset to IBM and OS/2, and appointing him to his position was something that you can definitively say that IBM did right.
IBM’s setting up of the IVL and various related marketing efforts somewhat balanced out its decision to close down the Desktop Software division, but did not make up for losing the applications that would have otherwise launched with OS/2 2.0 and helped it gain traction. IBM was helped out however by the fact that Microsoft and its legendarily cutthroat tactics was starting to loom large and menacing in the eyes of a number of software developers, who were beginning to think that maybe IBM was the lesser of two evils.
Taligent
IBM also diverted a lot of wasted effort into an abortive partnership with a directionless Apple, with both companies jointly working on a project called Taligent. Taligent was supposed to be a next generation operating system, one that Apple badly needed but seemed a bit of a head scratcher for IBM to be involved in. This description actually doesn’t do justice to just how odd Taligent was, as In Search of Stupidity puts it, Taligent started out as an attempt to build yet another next-generation OS, which then morphed into a half-witted effort to build an OS that would run other OSs. When this proved unfeasible, Taligent decided to waste more time and money creating a series of middleware tools that no one understood or bought before someone woke up and pulled the plug on the entire fiasco.
Before it finally imploded, Taligent burned through about half a billion dollars, although I am unsure if that was its total cost or just what IBM spent on it. Regardless, IBM definitely spent a lot of money and wasted a lot of effort that OS/2 badly needed spent on it instead. Thanks to Taligent and the Apple partnership, there were even rumors that Apple and IBM were going to merge, and OS/2 was going to replace its existing GUI with the Macintosh one. Fortunately for IBM, most OS/2 developers seemed to ignore this rumor and kept their efforts focused on selling and developing their OS/2 applications.
Of course IBM was busy making their job much harder than it needed to be, since IBM still refused to set up direct marketing efforts or distribution channel programs aimed at increasing the visibility of OS/2 applications from third party developers. As Chapman puts it: “Several attempts were made to convince the powers that were to create software promotional bundles with OS/2, or at the very least include trialware versions of applications in retail units of the product. All such attempts foundered.”
Something like this not could have not only helped to directly increase the sales of the third party applications that OS/2 desperately needed, but also encouraged third party developers to keep working on OS/2 applications instead of feeling like they were completely alone and without any support in selling their products.
About all you could say about IBM’s attitude towards its third party OS/2 developers was that it was basically benign neglect, not active hatred. I guess if you put IBM’s behavior next to the hostility Atari had originally had towards any third party developer that dared to develop games for the Atari 2600…IBM looks fairly good. But absent such a favorable comparison, IBM’s behavior makes it quite hard to feel sorry for what eventually happened to them.
OS/2 2.x
And IBM’s complete lack of intelligence in handling their third party developers pales in comparison to mind numbingly foolish decision to incorporate Windows 3.0 into OS/2 2.0 and Windows 3.1 into the following release, OS/2 2.1. Keep in mind, at this point Windows was rapidly gaining traction and market share, but was still very much in the minority when it came to its install base, not because of OS/2 or any other alternative operating system, but because most PC’s still solely ran MS-DOS, without any sort of graphical user operating system.
And with that in mind, let’s also talk about MS-DOS compatibility in OS/2 before tackling its Windows compatibility. After all, Windows might be shiny and finally selling really well, but MS-DOS applications were still a huge part of the market and OS/2 had excellent MS-DOS compatibility, with the OS/2 2.x line seriously beefing up what it could do. While the OS/2 1.x line could only run a single DOS session in a single window at a time, the 2.x enabled the user to run multiple DOS sessions simultaneously, each in their own window and virtual machine.
There was no need for IBM to include a full version of Windows with OS/2, it served no useful long-term strategic purpose and in fact would serve as a sort of trojan horse that would contribute to OS/2’s failure. However thanks to the remnants of the agreement with Microsoft, IBM was free to include a full version of Windows with OS/2, where it ran inside of a virtual machine.
And by “free” to do so, I mean that IBM had the right to do it, but also had to pay Microsoft for each copy of OS/2 sold. Byte magazine in 1994 stated that the license deal required IBM to pay twenty dollars per copy of OS/2, something that contributed at least 50 million dollars to Microsoft’s war chest from the OS/2 2.0 sales alone and helped contribute to OS/2’s high price of over 200 dollars. Although a twenty dollar royalty fee definitely does not excuse charging OS/2’s far higher price than Windows 3 and 3.1 did. But poor pricing for OS/2 is a bit of a recurring theme for Big Blue and it's going to come up again.
Another word on OS2 2.0 and 2.1’s Windows compatibility. While it was quite good, there were still issues that could arise due to memory management conflicts. Basically, Windows 3 and 3.1 have their own memory manager, as does OS/2. In order for the OS/2 2.x series to run Windows 3 and 3.1, the two clashing memory management systems had to be forced to work together somehow. It seems that this was never fully satisfactorily worked out, as Byte magazine said in the January 1994 issue that “Unable to modify the Windows code to use OS/2’s memory management services directly, the OS/2 developers settled on using the Windows memory manager within the OS/2 memory manager. Windows’ manipulations of memory can spill over into the OS/2 swap file.”
A quick caveat on OS/2.0’s Windows compatibility. While the vast majority of Windows and DOS applications ran without issue, it was unable to run any Windows applications that ran under Enhanced Mode. This represented a minority of applications, but did prevent OS/2 from being able to legitimately claim full Windows compatibility.
IBM’s marketing positioned OS/2 as able to run the vast majority of DOS and Windows applications, as well as its own native applications of course. And Byte credited this compatibility with OS/2 2.0’s sales numbers of at least 2.5 million. But Windows compatibility was a two edged sword.
The fact that Windows’ applications ran reasonably well under OS/2 created a significant positioning problem for OS/2. Because Windows was exploding in popularity, while OS/2 was still struggling to gain traction, a number of developers who had committed to creating OS/2 applications instead targeted Windows, and then used the loophole of OS/2’s Windows compatibility to claim that their applications were, in fact, OS/2 compatible. What IBM had done was basically take their enemy’s trojan horse, cheerfully invite it into their city, and then brag in their marketing about doing so.
Sadly IBM’s marketing of OS/2 was consistently either wrongheaded, poorly implemented, or just baffling. No better example of this can be found than IBM’s mind-numbingly pointless decision to sponsor the Austin, Texas based college football Fiesta Bowl from 1993 to 1995. This sponsorship, which cost IBM millions of dollars, made absolutely no sense from the start.
Who exactly were they reaching? As In Search of Stupidity puts it “it was unclear what benefit IBM derived from slapping the name “OS/2” on a second-tier sporting event. No demographic information seemed to exist that indicated that people who watched the Fiesta Bowl were also highly interested in 32-bit OSs, and there wasn’t much proof that watching a college football game would make people more inclined to rush home and demand computer resellers stock up on OS/2.”
Once the sponsorship went through, the event was renamed to the “IBM OS/2 Fiesta Bowl” at which point IBM discovered that it was its prerogative-and more importantly its responsibility- to provide a line-up of third parties who would be advertising during a series of time slots that came with the sponsorship. This was news to IBM, who apparently had not read the fine print of the contract. Perhaps their lawyers were off sick that day.
This could have been an opportunity for IBM to boost some of the best OS/2 applications and third party companies, specific applications and their features always being an easier way to grab a potential customer’s attention as opposed to touting the benefits of the operating system they run under.
I mean, it still wouldn’t have redeemed the inherent stupidity of sponsoring the Fiesta Bowl in the first place, but it would have at least possessed the talking point of quasi-coherence as a making-lemonade-out-lemons marketing strategy.
OS/2 2.1
Additionally, May of 1993 would see the release of the amazingly creatively named OS/2 2.1, which was a significant upgrade, more so than the slight version number increase would indicate. I’ll let IBM spokesman John Soyring give a quick summary of the changes between 2.0 and 2.1, especially the increased Windows compatibility and the friendlier, more approachable object oriented desktop. Now I am not buying his claim that OS/2 was easier to use than the Mac OS, which for all of its problems in 1993, remained the gold standard in approachability and ease of use. And I’m sure that my comments section will probably have some differing opinions on this as well, but this is my opinion and I’m sticking to it.
However the features he brings up, especially the improved DOS and Windows compatibility, for all the fact that I think it was a mistake to include it, were definitely attention grabbing. A stable, easy to use, powerful multitasking operating system that could run all of your existing Windows and DOS applications, do it seamlessly, and do it on your ordinary 386 or 486 consumer desktop, backed by the power and resources of IBM…well that would seem to be a pretty good value proposition.
An advertising campaign that played these facts up, and then got people excited about a bunch of spiffy new OS/2 applications would seem to be at least an effort in the right direction of building some hype. The broader consumer market at that point really didn’t buy an OS just to have it, so mentioning the new OS/2 release and then quickly pivoting to some exciting new applications to try to build some anticipation would have at least made some sense. And IBM, as previously mentioned, had a bunch of newly acquired ad space at the Fiesta Bowl.
However IBM of course…made zero effort to make use of the ad space for OS/2 applications. Rather than quickly calling up their best OS/2 application developers and giving them the opportunity to do a quick commercial for their product, IBM instead desperately reached out to local Austin businesses to see if they wanted the time slots. Advertising from barbecue joints, garages, and car dealerships would of course contribute absolutely nothing to OS/2’s struggles, but such a thought apparently never occurred to IBM.
The end result of this exercise in um…creative…marketing was not successful by any metric known to man, and within IBM the Fiesta Bowl became quietly known as the Fiasco Bowl.
As previously mentioned, OS/2 2.1 was a significant upgrade for OS/2 users. As a matter of fact it was awarded a Byte Award for Excellence in January of 1994, in Byte magazine’s annual roundup of the best products. In the article explaining the reasons for the award, Byte declared that “The latest revamp of OS/2 has solved most, if not all, of the problems plaguing the long-in-coming version 2.0.” and then went on to laud OS/2 as “an operating system that is finally winning some of the vital support it need, and, as of this version, truly deserves.” Which was all definitely good news for OS/2 and boded well for its sales.
Furthermore, the article also contained a further quote from Byte’s executive editor Jon Udell, saying “Running OS/2 2.1 on an 8-MB machine that hasn’t a prayer of running Windows NT, I find that it delivers many of the same benefits: robust multithreading and multitasking, a comprehensive 32-bit API, an advanced file system, and competent support for Windows 3.1 software.Software developers have known for years that OS/2 is a far more productive environment than DOS plus Windows. With the polish and maturity of version 2.1, more and more users are discovering the same advantage.”
This was a big win for OS/2, especially being favorably compared feature-wise with the shiny new Windows NT, but with lower system specs. Lower required system specs from a Microsoft competitor is hardly new of course, but the thought of that coming from IBM, the company that Microsoft programmers had long sneered at for being stodgy, not hiring the best people, and for having a focus on lines of code rather than efficiency…well that was a bit of a surprise. But OS/2 2.1 was proving that IBM’s programmers…had learned at least a thing or two about code optimization and tighter programming.
And there was further good news from IBM’s development efforts, with the November 1993 release of OS/2 for Windows, the problem of paying Microsoft royalties for including Windows with every OS/2 sale had been at least partially solved via updating OS/2’s Windows support to allow for a user who already owned Windows 3.1 to install OS/2 on top of it, and OS/2 would then use the user’s pre-existing Windows install to drive its Windows compatibility.
Hence, OS/2 for Windows didn’t need to include a copy of Windows in every box and I guess the theory was that sales reps would point people in the direction of the correct OS/2 2.1 version for their needs, whether with or without Windows. However I did find an article complaining about the poor service OS/2 received from sales reps, so I suspect this may have created a significant amount of confusion. IBM would get a mild amount of revenge the following decade when Microsoft decided that what people really wanted was to choose between approximately forty-two versions of Vista, some of which sold better than others of course.
Additionally, from what I can tell, IBM still had to pay a royalty to Microsoft for every version of the normal OS/2 2.1 sold, as that included Windows in it, it just didn’t need to pay royalties for any copies of OS/2 2.1 For Windows sold. The lack of a need to pay royalty fees meant that OS/2 for Windows actually only cost half of what the normal version of OS/2 2.1 cost.
OS/2 Warp 3
In spite of IBM’s lamentable tendency to repeatedly shoot themselves in the foot and then look confused, OS/2 somehow actually continued to grow in popularity and its user base was steadily increasing. This was definitely due more to its growing reputation for stability and ease of use and the good word of mouth generated by contented users and the increasingly fervent OS/2 fanbase than it was from IBM’s continued incompetent marketing department. The growing concerns about Microsoft’s behavior and business practices didn’t hurt either, with IBM increasingly benefiting from its reputation as stodgy and out of touch, rather than pure evil incarnate.
However, in October of 1994, OS/2 3.0, was released to the public. It would be followed by an updated version, OS/2 Warp with WinOS2, scheduled for release in early 1995. The versions were differentiated by their support for Windows 3.1, with the original Warp release, OS/2 Red lacking built in support for Windows 3.1 applications, and the later release, OS/2 Blue, coming with a full version of Windows 3.1 preinstalled, that ran under a virtual machine. These names are not official names, but came about because of the two different spine colors on the OS/2 retail boxes.
This might come off a little confusing so let me clarify: either version of OS/2 could run Windows, but OS/2 Red didn’t include it so the user needed to have it already installed on their system, and then install OS/2 on top of it. OS/2 Blue came with a complete copy of Windows 3.1 built into it, and thus could be installed directly without needing anything else in order to run Windows 3 era applications. This was undoubtedly done for the same reason as OS/2 2.1 had done it: to reduce the amount of royalties that IBM had to pay to Microsoft in order to license Windows. Of course it did rely on the sales rep being knowledgeable enough to point the customer to the correct colored spine for their needs.
Either way it was done, OS/2’s Windows compatibility was extremely easy to use, as from the user’s perspective all that happened was a new window popping open that was running Windows 3.1, complete with Windows 3.1’s Program Manager. Any 16 bit application that ran under Windows 3.1 would run just fine, provided the system resources were sufficient.
And this release came with a marketing failure so complete that it almost killed OS/2 all on its own. And it all started with Star Trek, Paramount, IBM chairman Lou Gerstner, and some remarkably incompetent performances from IBM’s lawyers, who were apparently seeking to prove that marketing wasn’t the only division of Big Blue that could royally screw things up. For years OS/2 releases had been given internal names lifted from Star Trek, including “Borg” and “Klingon”. To be very clear, these names were merely internal code names, not names for the actual OS/2 releases. In keeping with this tradition, OS/2 3.0 had been given the internal code name of Warp. Unsurprisingly this was in reference to “warp speed”, a fitting term for OS/2 3.0 as it was supposed to be noticeably faster than previous versions.
As its release date neared, Chairman Gerstner decreed that Warp would be the official name of not just this release, but of the product as a whole, with OS/2 being henceforth known as “OS/2 Warp” and this release being “OS/2 Warp 3”. The Star Trek connections with this name were cool, hip, trendy, and many other spiffy adjectives that IBM had never been associated with, so it seemed like a smart marketing coup. IBM went ahead and based their entire marketing for OS/2 3.0 around a fairly obvious Star Trek based theme, renting out a big hall in New York City for the announcement, sending out hundreds of invites to various movers and shakers in the computer world to come cover the announcement, and for a special treat, invited Patrick Stewart to come headline the event.
It was going to be incredibly splashy, really giving OS/2 Warp a bold entrance into the marketplace and hopefully associating it in people’s minds with the aforementioned cool adjectives. And with OS/2’s maturing as a product, it was a fairly worthy target for such branding. OS/2 Warp was a very stable operating system, it maintained its excellent backwards compatibility with Windows 3 era applications, and of course it could run the vast catalog of MS-DOS applications that were still a significant part of the software market. It was also fully 32 bit, in an era when consumer Windows was still struggling to awkwardly straddle that gap. Windows NT is another topic of course, something we will be dealing with in the future in a separate video.
And the OS/2 world was very hopeful about this release and IBM’s plans for handling it and marketing it to the mainstream consumer. It seemed like IBM finally had an unqualified winner on its hands from both a feature and a marketing perspective. The OS/2 faithful were optimistic that this time, Big Blue was going to do right by their favorite operating system. And the story of how it all went horribly wrong, and the faithful OS/2 supporters gathered in NYC didn’t even get to see Patrick Stewart…is just painful.
OS/2 Publications
A quick word here on OS/2 specific publications by the way, as this will become relevant in a minute. Apart from IBM’s own OS/2 Developer magazine, there were apparently at least three major outside OS/2 publications, OS/2 Professional, OS/2 World, and the descriptively named OS/2 Magazine. Actually there were two different publications named OS/2 Magazine but one was apparently renamed to OS/2 and Windows Magazine just a few months after it launched in 1990 so I am not counting it. All were independent of IBM, but as previously mentioned at least the first two were formed thanks to the efforts of IBM’s Independent Vendor League.
From what I gathered, IBM ensured that at least the first two of these magazines had access to IBM’s registered user list, gaining an immediate audience of 70,000 recipients for the inaugural issue of OS/2 Professional in November of 1992 and presumably a similar one for OS/2 World. I don’t know very much about OS/2 World and I have not been able to find an archive of back issues, but OS/2 Professional is an excellent resource for tracking the ups and downs, wins and losses of OS/2 at any given time during the magazine’s run.
I have no idea if publisher and Editor-In-Chief Edwin Black ever heard this, so on the extremely off chance that he ever sees my video, let me just say that you and your staff at OS/2 Professional did a great job covering OS/2, warts and all and your hard work was very helpful in creating this video.
And Edwin Black is another individual that I will be periodically referencing, as he is a great example of someone outside of IBM, who wanted OS/2 to succeed and was extremely passionate about it. His magazine, OS/2 Professional is a great read, and clearly shows the love and passion that motivated a lot of the OS/2 community. It also is a great way to see a number of examples of frustration that boiled to the surface periodically when the fans of OS/2 saw IBM once again failing to handle OS/23 correctly.
IBM’s Personal Systems Programming, or PSP division contributed a letter to OS/2 Professional’s inaugural issue, saying On behalf of everyone at IBM Personal Systems Programming, congratulations on the debut of OS/2 Professional. We're pleased that your publication is dedicated to serving the needs of the rapidly expanding community of users of OS/2 2.0, IBM’s advanced 32-bit operating system. While many in the industry talk about the operating system of the future, IBM has already shipped more than one million copies of OS/2 2.0. In the coming months, we hope OS/2 Professional will become an important source of news and information as we add new products and functionality to OS/2, and users find new ways to increase their productivity. Thank you for joining us as we build the bridge between yesterday and tomorrow with OS/2 2.0.
Fine words, even if they are pretty much just bland corporate speak, but their source being IBM’s PSP division is rather ironic in light of just how much damage this division was going to do to OS/2 in the coming years. Damage so great that OS/2 Professional itself was going to be explicitly calling for this division to be closed down in a May 1995 publisher’s memo from Edwin Black that was very bluntly called “Dismantle PSP”. But that was a couple years and some considerable bad blood away.
OS/2’s Professional’s editorial bent was of course very pro-OS/2, as might be expected, but it took a very honest and straightforward position editorially, even though it obviously depended heavily on the revenue gained from the ads placed by IBM itself, especially at first. This meant that while the magazine would get excited about OS/2 news and hype, it would also call out IBM for shortcomings and especially its botched marketing. And let's just say that IBM’s handling of criticism does not exactly show them in a good light.
Articles that covered both the good and the bad of OS/2 were apparently extremely negatively received by IBM, who seems to have wanted nothing but praise and adulation from extremely compliant publications. Or in other words, IBM wanted Nintendo Power but instead got Next Generation Gaming.
I mean, just watch this segment from a 1993 episode of The Computer Chronicles, called “What Happened to OS/2?” where Black and OS/2 Professional are introduced. Does this sound like a man who was anything other than passionate about OS/2 and enthusiastic about its features? This is hardly the behavior of a man who wanted anything less than total victory for OS/2. Yet it seems that his willingness to call out IBM’s many failures with OS/2 didn’t exactly endear himself to Big Blue.
While I only have the magazine’s word for it, the evidence seems pretty conclusive that IBM continually displayed a very, very thin skin in regards to any editorial criticism. In a June 1995 article called Caught Between Microsoft and IBM, Black wrote that the magazine’s previous coverage of OS/2’s failures had provoked IBM’s ire such that, in Black’s words: “In the past, IBM has harassed me with telephone calls to my house at 1 AM to complain about articles, instructed its Canadian ad agency to pull CSet advertising and explicitly state it was in retaliation for articles, faxed our confidential business plans to competing magazine publishers and threatened to pull our license to use “OS/2” in our name unless we ran a photograph of a PSP marketing executive—and each time we have steadfastly refused to cave in.”
And having read a decent amount of OS/2 Professional’s Warp coverage, I have to say that they come across very even handedly, clearly people who loved OS/2 and were big fans of it but also weren't blind to its issues and weren’t afraid to call out IBM’s poor decisions, especially in the area of marketing. IBM, flailing around on all sides against the consequences of its poor decisions, was apparently only willing to listen to unqualified praise of OS/2 rather than “tough love” by the people who honestly seemed to love OS/2, hate Microsoft, and want OS/2 to succeed with the very best product possible, backed by smart marketing. It’s fairly easy to draw a straight line from this attitude to OS/2’s eventual fate.
OS/2 Warp 3 Coverage
So with all the aforementioned in mind, let's take a look at what OS/2 Professional had to say about OS/2 Warp 3. In the October 1994 edition, a publisher’s memo to the reader called “Warp Launches” appeared before the issue's two main features on Warp, both of which were drawn from using Warp’s final beta.
This memo strikes a painfully optimistic note about IBM’s OS/2 Warp plans, stating “IBM worldwide is pulling out every marketing stop it has. Armonk is involved as never before. Lou Gerstner personally renamed the product OS/2 Warp just days before launch. And this time, IBM is putting its money where its mouth is: TV, radio, billboards, direct mail, publications, retail programs, OEM efforts. And it's all happening simultaneously around the world.”
The publisher’s memo went on to state that the research that went into the issue’s coverage of OS/2 Warp was: “the most fast-paced—and frustrating—assignment in my 26 years of journalism. It took three times as long as my investigation of the Kennedy Assassination that helped reopen the Warren Commission; three times as long as an investigation of Sen. Dave Durenberger that led to his indictment; twice as long as the opening article on the largest malpractice insurance fraud in the history of the U.S., resulting in numerous prosecutions. And it took 12 times longer than a similar story on Computer Associates in OS/2 Professional last year. What normally takes one to two calls for any other story, took 12- 15 calls to nail down the details of OS/2 Warp Version 3.”
The two main features on Warp in this issue are grouped under the overall title of OS/2: The Next Generation. The first feature covers OS/2 from an installation and general use standpoint and is so blatant about drawing connections to Star Trek that the article title is Working at Warp Speed and there is an actual image of the Enterprise on page 1. As a matter of fact, the Enterprise is part of both features on Warp in the magazine, and Star Trek is an implicit part of every title.
The other article, which covers the marketing plans for OS/2 Warp, is called The Warp Campaign: Phasers on Stun. This sort of shall we say “Star Trek adjacent” marketing was about to bite IBM rather painfully in their suit clad hindquarters, but that’s getting ahead of the story.
OS/2 3 Warp was shaping up feature-wise to be a great product, and one that IBM badly needed, but the review team at OS/2 Professional was not going to just declare it perfect and go home. Errors in the install process are noted, including one that showed up when attempting to install OS/2 3 Warp over a previous version of OS/2 that required over half a dozen wipe-and-reload-from-backup before the correct solution was found. This solution required manually editing CONFIG.SYS before doing the upgrade, always a fun part of any operating system upgrade back in the day. In fairness the article points out that they were working off of a beta release of Warp, and that these bugs had most likely been resolved before the final release.
Jerry Pournelle contributed an article as well, one that was less laudatory towards older versions of OS/2. Basically he took issue with how pre-Warp versions of OS/2 could put the user in a bind where the Windows virtual machine locked up so badly as a result of a misbehaving application that the user couldn’t get back to OS/2 itself, forcing a whole reboot and a potential loss of work.
To be fair, I am not exactly sure of the relevance of this article, given that Warp fixed the major issues Pournelle was writing about, however I like the article’s inclusion as it shows that the magazine was doing its best to balance being a cheerleader for OS/2 with being even handed in its coverage of it. Plus I am always entertained by Pournelle’s typical article formula of breaking something while trying to do something, and the detailed steps taken to resolve the problem.
The overall impression of OS/2 3 Warp was very good however, with the review pointing out among other things that “OS/2 Warp will now run reasonably well on a 4mb system, with a realistic “sweet spot” of 8mb or more—a decrease from 2. l’s 8/16mb target”. You heard right, IBM’s fancy new version of OS/2 actually had lower system requirements than the previous version. This might be the first time this ever happened, and it certainly was a worthy achievement, one that Microsoft still hasn’t managed to pull off.
OS/2 Warp was additionally lauded in the review for its vastly improved multitasking, making it far harder for a misbehaving application to lock up the entire computer and force a restart. However Warp did have a rather serious issue with the GUI itself. According to an excellent article from Ars Technica on OS/2, it was possible for a misbehaving OS/2 application to lock up the entire OS/2 GUI. To be clear, the computer was still running and background processes were still executing, you just couldn’t, you know, interact with anything.
In the words of the article “Unfortunately, OS/2 had a crucial flaw in its design: a Synchronous Input Queue (SIQ). What this meant was that all messages to the GUI window server went through a single tollbooth. If any OS/2 native GUI app ever stopped servicing its window messages, the entire GUI would get stuck and the system froze. OK, technically the operating system was still running. Background tasks continued to execute just fine. You just couldn’t see them or interact with them or do anything, because the entire GUI was hung.”
This was definitely a problem, although in fairness the computing world of the mid-1990s was far more tolerant of computers freezing and requiring restarts than we are today. Thanks to the cooperative multitasking of the classic Mac OS, I clearly remember needing to reboot the family Performa fairly frequently whenever Warcraft II or Starcraft decided to be naughty and freeze everything when switching between full screen and windowed mode. So Warp’s issue with occasionally freezing wasn’t as big a deal as it would be seen today, but it definitely was a negative.
On the positive side, OS/2 Warp was finally shipping with what was called a BonusPak in every box. This was an application suite that included a spreadsheet, a word processor, a flat file database, and a personal information manager. Most of these applications had been available before, but bundling them together and including them for free in every boxed copy of OS/2 sold was an unusually smart move for Big Blue. And these applications were solid enough in their features for most users, with OS/2 Professional saying “The applications within the BonusPak are robust enough for most users, and include such advanced features as mail merge, DDE links between data in the disparate applications, spell checking, and data import and export.”
Overall, the OS/2 coverage in the magazine was thorough, fair, and about the farthest thing from a corporate snow job one could imagine. It also relied on a rather heavy dose of optimism about IBM’s competence, and some equally optimistic assumptions about problems in the betas that would surely be cleared up before Warp went gold and the final preview copies went out. In fairness I must note that IBM at least gave a fair impression of doing a lot right during the run-up to Warp’s full release, and I don’t see the magazine’s optimism in any way as somebody closing their eyes, sticking their fingers in their ears, and hoping, rather it was optimism that seemed to at least be reasonably warranted.
And I also didn’t quote or cover any of this for humor at anyone’s expense, rather I want to very clearly make the point that the OS/2 community loved OS/2 in a way that IBM, bland corporate master of staid business machines that filled entire rooms, never really understood and appreciated. There was genuine fervor and devotion to OS/2, and OS/2 Warp was generating a lot of well deserved excitement and hype in the community. Warp seemed like a fit product to go to warp speed, and to the ranks of fervent OS/2 supporters, the future seemed bright.
Now let’s return to IBM’s planned exciting release of OS/2 Warp and the event it was throwing in New York City to launch it, and its Star Trek adjacent branding. All seemed good to go as the date for the event drew near, and the OS/2 faithful got more and more excited. This was going to be a memorable night. Maybe even as exciting as the iconic launch of the Macintosh a decade prior. Okay definitely not that exciting, but still, the light coming off of OS/2 Warp was almost blinding.
The Marketing Disaster
And then a rather significant problem abruptly appeared on the horizon…and suddenly slammed right into IBM. In all of the excitement IBM had somehow neglected to reach out to Paramount, the owners of the Star Trek brand and franchise, and make sure that they were fine with IBM making such an explicit connection with the Star Trek brand, a connection that had already been widely embraced by the OS/2 community, as is evidenced by the aforementioned OS/2 Professional article.
Paramount, who had already gotten annoyed at the casual way IBM had publicly used its internal Star Trek code names for previous versions of OS/2, decided to cut IBM off at the knees. Official legal notices were sent from Paramount’s lawyers to their IBM counterparts, warning them that “Warp” was a trademark of Paramount, and could not be used in any way that tapped into the Star Trek brand, whether implicitly or explicitly. IBM could use an alternate dictionary definition of “warp”, but not in any way that infringed upon or suggested the Star Trek brand.
This was…a bit of a problem. Actually a huge problem, especially from a marketing perspective. IBM had already put a lot of time into establishing the name OS/2 Warp in people’s minds, and it was felt that it was too late to change to another name. Regretfully, nobody suggested reaching out to George Lucas to see if he could be persuaded to let IBM rename OS/2 Warp to OS/2 Hyperspace or OS/2 Millenium Falcon. Sadly, instead of that, IBM’s marketing department decided to reach for new heights of umm…creativity.,..and stick with the strict dictionary definitions of “warp”, such as “bent”, “twisted”, and “warped”. Oh and adding insult to injury, Patrick Stewart did not appear at the Warp launch as originally planned, instead the OS/2 faithful had to settle for Kate Mulgrew, from the then upcoming Star Trek Voyager show.
The desperate attempt to redo the entirety of OS/2 Warp’s branding and theme while still keeping the same name resulted in a psychedelic advertising campaign that, to an overly critical mind, may have been dreamed up from the center of a That 70s Show camera twirl. Advertising was created that encouraged the prospective user to “warp” their computers by upgrading to OS/2. Channeling the spirit of Cheech and Chong for an operating system that already had an uphill battle was a…unique way of selling to the mass market of the mid-1990s. As In Search of Stupidity puts it, “Everyone, of course, was thrilled at the prospect of running a psychedelic, warping OS that smoked dope and had flashbacks when you asked it to retrieve a file.”
And the full rollout of Warp was also further marred by the fact that on top of struggling to reposition OS/2 Warp’s branding away from Star Trek at the last minute, IBM also had to deal with getting its new marketing agency, Ogilvy & Mather, up to speed. And why was this necessary? Because this agency had been brought in to replace IBM’s previous marketing agency, Lintas, who had been fired in May of 1994.
When fired, apparently one of the Lintas executives rather presciently stated: “They’ll never be ready for the fall.”, meaning the October release date for Warp. So Ogilvy & Mather basically got dropped into the thick of things at the last minute, and had to struggle to get the revised marketing for Warp sorted out and running.
Unsurprisingly, the turmoil generated by trying to change up ad campaigns at the last minute, plus deal with the change of marketing agencies, plus IBM’s own unique brand of bad luck/poor planning…well it was just too much. As a matter of fact, it seems that IBM’s advertising executives in the PSP division had no idea who their Ogilvy & Mather counterparts were until…about a month before Warp’s October release.
This obviously made the already difficult challenge of marketing Warp even harder. It seems that literally everything that could go wrong from a marketing campaign…did go wrong for poor Warp before it was even out the door. The result of all of this hasty scrambling around was an ad campaign known as “OS/2 Warped” and it was every bit as confusing and poorly thought out as you would expect by this point.
OS/2 Professional wound up doing a full article/post mortum on the marketing disaster and they pulled absolutely zero punches, stating that “The launch itself, which even this reporter hailed as a great event, was much ado about nothing in the wake of an advertising campaign that did not get it right, a marketing strategy that got it wrong, and a product that unfortunately made headlines for its boo-boos.” Ouch.
I will add that it seems like the idea for going with the traditional meaning of “warp” apparently came from and was implemented by Ogilvy & Mather, so while IBM signed off on it and thus bears its fair share of blame, the failure was a team effort. Although fairness also compels me to balance this out further by pointing out that the decision to stay with the Warp branding but use an alternative meaning definitely came from IBM, and thus Ogilvy & Mather were pretty limited in what they could do with their limited time to plan.
As the OPS/2 Professional post mortem says in a great paragraph that really encapsulates the entire problem with the whole “warped” campaign: “Most significantly, O&M used the wrong Warp concept. Gerstner didn’t want to reinvent the well-established Warp name used throughout the beta period. But that name, Warp, was taken from the Star Trek lexicon, evoking images of fantastic speed and physics-boggling abilities. O&M and PSP opted for the distorted, wobbly, can’t-get-straight Warp, the misshapen wood Warp, the twangy guitar Warp. The wacko, wobbly, off-key Warp image O&M propagated did not have the same intrinsic power of the “faster than light” Warp. Voila, the power of the Warp name IBM invested so heavily in, the one Gerstner sought to capitalize on, was in fact lost, or worse, turned into a negative image business could not have confidence in.”
OS/2’s baffling ads for getting “warped” were not exactly setting the faithful OS/2 fans on fire with enthusiasm. Quite the opposite actually, as the publisher of OS/2 Professional, Edwin Black, wrote a steaming editorial for the magazine in which he expressed his extreme displeasure with an OS/2 ad he had seen while passing through O’Hare airport. In this ad, legendary basketball coach Phil Jackson smiled dreamily at the thought of warping his computer.
OS/2 Warp also had a series of ads that never actually showed OS/2 at all, rather it was just shots of people’s faces reacting to OS/2 Warp’s various features. The ad campaign was called “It’s a Warped World” and the ads ended with the painfully “hello fellow kids” phrase “OS/2 Warp, a totally cool way to run your computer.”
Still, OS/2 Warp somehow soldiered on, helped along not only by its reputation for stability, but also by Microsoft’s increasingly bad reputation with the US Department of Justice. More developers were creating OS/2 applications, not just rebranding their Windows ones via using the fig leaf of OS/2’s Windows’ compatibility. Retail sales of OS/2 itself began to pick up strongly, and at the 1994 Technical Interchange trade show, an IBM run event, a large number of software vendors had completely sold out of all of the OS/2 applications they had brought as eager passers-by snapped them up.
The ability to easily run the vast library of 16 bit Windows software inside of a true 32 bit system that also featured proper preemptive multitasking was a huge selling point for driving sales…if properly marketed. Although a caveat on this compatibility is the fact running virtual machines on the slow consumer hardware of the mid 1990s may or may not have provided a good experience on demanding Windows applications. I remember trying out SoftWindows 98 on our family iMac and it wasn’t exactly a speed demon, although of course Windows was far better integrated into OS/2 than SoftWindows was on Mac OS Classic.
Some possible good news was the fact that Windows 95’s beta had shipped with a nasty bug that caused it to freeze when attempting to run multiple 32 bit applications, something that Warp handled well. This bug was called out from the front page of Infoworld, in their March 27th, 1995 edition, with the cover title reading “Operating System Freezes Running 32-bit Multitasking Apps” and a money quote inside that said “What was publicized as the largest beta program in history failed to turn up a fundamental architectural flaw in Windows 95 that causes the operating system to freeze when multitasking a 32-bit application...".
This could have been turned into excellent marketing for Warp as well, helping it to build as much market share as possible before Windows 95’s actual release, and finally giving IBM the firm foothold in the operating systems market that they had been groping their way towards for years. There was certainly no way to stop the oncoming juggernaut, but favorable press around a good product given months of unchallenged time to build sales…well it goes without saying that this was the best opportunity IBM had been given since the mid-80s.
Some parallels may be drawn with Sega’s attempt to build as big a market share as possible for the Dreamcast before the PS2 launched. And if you know how that attempt wound up working out…well nobody watching this is doing it from their Dreamcast 3, in between rounds of Modern Warfare: Panzer Dragoon 4.
In May 1995 an mildly updated version of OS/2 Warp aimed at the enterprise market was released, called OS/2 Warp Connect. A New York Times article from that same month stated that this version was designed to make it easy for thousands of computers to share not just hardware such as printers, but also various applications and information, saying “The intention is to provide a single product that lets customers share printers, software applications and information with a variety of computers and operating systems -- without having to spend hours learning how to interact with each one.”
That same article goes on to say that OS/2 had sold nine million total copies of OS/2 in total, two million of which were copies of Warp that had been sold since its introduction in fall of 1994. That same article does unfavorably contrast this number with the sixty million copies of Windows that had officially sold, which of course does not include the rampant piracy of early versions of Windows that Microsoft, ever aware of the advantages of market domination, was happy to turn somewhat of a blind eye towards.
Of course OS/2 was still in no danger of supplanting Windows at this point, not with its existing market share plus the imminent release of the juggernaut of Windows 95 just around the corner, but IBM was still settling into a decent, if distant second place, as opposed to Apple’s rapidly dwindling market share at this point. Warp was on the verge of taking OS/2 to sustainable success, and in spite of IBM’s best efforts, things seemed to be breaking its way very favorably.
It would take a serious act of incompetence to throw this away, something genuinely, monumentally stupid. As usual, IBM was up to the challenge.
But first let’s set the stage a bit with another monumental piece of poor decision making by Big Blue…who you might remember also had a personal computer division. A personal computer division that even though it wasn’t the largest PC maker anymore, still was a major player in the PC market. You would think that IBM would have started pre-installing OS/2 on every IBM PC that went out the metaphorical door just as soon as it was possible. You would be wrong, as IBM’s hardware and software divisions apparently enjoyed treating each other like bitter enemies.
According to an excellent 2002 article called “Who Killed OS/2?” which you can read over at os2world.com: “the IBM PC Company and the IBM Personal Software Products division had this in common: They both sent their profits to Armonk, and if there were no profits, the firings were decreed in Armonk. There was no love lost between them. Each looked upon the other as the enemy -- PSP believing that the PC Company was trying to run them out of business by failing to preload their software, and the PC Company believing that PSP was trying to put them out of business by trying to force OS/2 on them when their customers wanted Windows. It was noticed that IBM apparently didn't think enough of OS/2 to put it on its own machines; a person wanting OS/2 on, for instance, an Aptiva would literally have to beg and demand and threaten. Then, when it all did arrive, it often as not wasn't properly installed and wouldn't run well. (Later, IBM would install both OS/2 and Windows on some of its machines, but made no particular effort to persuade the user to boot one over the other, and many customers were unaware that they had OS/2 at all.)”
And of course, since IBM wasn’t making any effort internally to include OS/2 as the default operating system on its own computers, it sure wasn’t putting in any effort to get OEM’s to include OS/2 instead of Windows as their default. Microsoft’s cutthroat willingness to do whatever it took to get an OEM to sign a Windows licensing deal, versus IBM’s at best lackadaisical attitude towards the same, ensured that Microsoft pretty much had the field to itself when it came to what operating system OEMs bundled with their PCs. Sure there were other companies, such as Digital Research, who were attempting to get OEM’s to bundle their product, but these were pretty small companies compared to the Microsoft behemoth, and Microsoft had little trouble squashing such attempts.
Clearly IBM was its own best enemy and had been for pretty much all of OS/2’s life, but the years of self-sabotage were about to come to a sudden and concussive end, and it was going to start from the very top, very publicly.
The OS/2 Implosion
On July 31st, 1995, IBM CEO Lou Gerstner gave a speech during his yearly meeting with financial analysts, where he not only said that OS/2 was waging the “last war”, implying that the current war was already lost, but he followed this up by saying that IBM had lost its opportunity to “go after the desktop.” It required very little spin for the news media to report this as IBM admitting that OS/2 had failed in the marketplace. The New York Times even ran an article on August 1 under the headline “IBM Chief Concedes OS/2 Has Lost Desktop War.” The fallout of this, unsurprisingly, was sudden and drastic.
OS/2 vendors began to publicly express doubts about the continued viability of developing OS/2 applications, a number of corporate installations of OS/2 were canceled, and major OS/2 columnists such as Will Zachman publicly announced their doubts about the platform. Will Zachman was a pretty big deal in the OS/2 world, as he ran the largest OS/2 forum on CompuServe, Canopus. So this would be roughly equivalent to, I dunno, CultOfMac or Apple Insider declaring that they had serious doubts about Apple and were considering alternatives.
Just five days after the first New York Times article detonated, a second article was run in their Technology Column section. This article was called “OS/2 No Longer at Home at Home” and it had lots of shocking quotes from IBM’s key OS/2 evangelist, David Barnes, who we met earlier. Some of the key quotes from the article were “OS/2 is a great operating system…Sony’s Betamax was a better system than VHS” and the real killer quote: “I’m going to put Windows 95 on the machines in my house.”
Reading through the article is pretty devastating to OS/2. Essentially the author, Peter Lewis, had been struggling to get the latest version of Warp to install on his new Pentium tower. After several days without success, Lewis called for help, and David Barnes was sent over to help him with his installation. I am assuming this occurred due to the potential bad PR that would occur if the New York Times published an article stating that it was very difficult to install Warp.
According to the article: “After spending more than an hour trying unsuccessfully to install OS/2 on my machine, the normally effervescent Mr. Barnes finally gave up. He remains the world’s biggest fan of OS/2 as a reliable operating system for big businesses. But would he recommend OS/2 to a friend who did not have a technical support team on call in the office? ‘Let me put it this way, Mr Barnes said. ‘I’m going to put Windows 95 on the machines in my house.”
And as previously mentioned, David Barnes was a big deal in the OS/2 community, having spent years traveling around to preach the virtues of OS/2. This was bad enough, but then it got even worse as IBM’s incredible ability to shoot itself in the foot manifested yet again.
First of all IBM did absolutely zilch for several weeks after Barnes’ comments were made public. Not one single word, even as OS/2 forums were blowing up with outrage and tech writers attempted to reach out to IBM for comments to no avail. The OS/2 exodus picked up more steam, as wavering users and businesses who had had their confidence shaken by Gerstner’s comments decided that they had heard enough. Eventually IBM realized that its strategy of playing some twisted version of dead possum wasn’t going to cut it. And so Big Blue decided to…eventually have Barnes release a mild PR statement that his comments had somehow been taken out of context and IBM spokescreatures pinky promised that OS/2 was still a priority for them.
Unsurprisingly, nobody cared at this point and OS/2’s demise was well and truly over and done with. I will say that OS/2 Professional, valiantly tried to apply some duct tape to the gaping hole that had been blown in the OS/2 ship, with a publisher’s memo from editor-in-chief Edwin Black in the October 1995 issue saying “Was it decent or professional of Peter Lewis of the New York Times to invite David Barnes to his home for some personal help, guy-to-guy, and then publish the off hand remarks that any professional journalist should know would be off the record. Shame on you, Peter. Your article was disgraceful and uncalled for. And IBM, being a cripple, refused to answer back.”
I’m not here to comment on the ethics of what happened one way or another, but I would imagine that it would be hard for a reporter to avoid the temptation to report on something like one of OS/2’s biggest supporters admitting that he was probably going to be running Windows 95 in his house when it came out.
Now, I do think that it’s very likely that there was at worst some hyperbole or spin going on here, and at best some nuance was lost when a private conversation went public, I honestly cannot say for sure and I am purely running on conjecture here. It’s highly possible that what Barnes actually meant was that Windows 95 was good enough that he was going to use it for certain things in his house that he couldn’t do with OS/2.
After all, OS/2 Warp was not Windows 95 compatible, it could only run Windows 3.0 and 3.1 applications, and Windows 95 was certainly going to have a lot of applications that wouldn’t be available on OS/2 and couldn’t be run through compatibility mode.
When Betamax and VHS were battling it out, there were certain movies that were only on VHS, and it wouldn't have been unreasonable for Sony employees to quietly have a VCR at home in order to watch these movies. You wouldn’t want to broadcast that if you were Sony, but it would be perfectly understandable, even if Betamax didn’t wind up eventually losing that battle.
But IBM’s lack of response gave rumor plenty of time to take flight and for the whole OS/2 community to get well on the way to melting down. By the time this whole situation started to, for lack of a better phrase, calm down, Windows 95 had already landed. Thanks to IBM’s botched handling of the issue, the only competitor it faced was the then imploding Apple. Oh and BeOS of course, can’t forget them. Side note, BeOS’s open source successor, Haiku-OS is a neat project that I hope continues to grow, and I am proud to say that I was one of the people who bought the Alpha version on DVD like ten years ago in order to help support the project.
Back in the imploding OS/2 world of October 1995, OS/2 Professional did its best to take an optimistic view of things and the future, saying in its publishers note that “OS/2 is alive and well. On the day Win95 shipped, OS/2 Express had its best day of the year. Every day thereafter, we enjoyed record setting days as our corporate and governmental customers made major multiple purchases. IBM products and OS/2 apps are selling better than ever. As I said last month, OS/2 will only become stronger following the backlash against Win95, which is already happening. The bottom line for all who care about OS/2—including me, is: don’t get angry, get even. Stick with OS/2.”
Again, OS/2 had a very devoted user base, one that I would say could match the Amiga user base for devotion, and one that did its best to be positive about the OS/2’s future, no matter how hard. You’ve got to respect that, and once again, the thought of such devotion accruing to an IBM product never ceases to be amazing. But it just wasn't going to be enough. So far as the broader consumer market was concerned, IBM itself had basically told them to never even consider OS/2. And so far as developers were concerned, IBM had also told them to never bother considering developing OS/2 applications.
And thus Windows 95 continued to sell in staggering numbers, finally definitively dethroning MS-DOS and starting to actually make a significant dent in the gaming market thanks to the first iteration of DirectX and the hard work of Alex St John and his team.
Yet OS/2 somehow, miraculously continued to stagger on, with its user base resting on a core of die-hard users, plus a number of institutions that had invested significantly in OS/2 installs or relied on OS/2 specific software. There was also one more major OS/2 release to come, plus IBM also had some rather odd stuff in the OS/2 pipeline that we will briefly summarize.
OS/2 Warp PowerPC
In December of 1995, a rather experimental version of OS/2 Warp was released for PowerPC, descriptively called OS/2 Warp PowerPC. Although this was right around the time that Apple was transitioning from the 68k architecture over to PowerPC, this version of OS/2 was not meant to run on Apple computers. Rather, it was aimed at a rather small line of IBM’s own line of PowerSeries desktops and Thinkpads.
OS/2 PowerPC appears to lie somewhere between beta software and an actual finished product, with some areas quite mature and stable, and other areas completely lacking. According to this website, it had almost no networking abilities at all, meaning no TCP-IP and no local networking, although it seemed to have included a way of accessing CompuServe via modem. A bit more detail on the reasons behind the missing network support comes from ibmfiles.com’s article, where it explains: “OS/2 Warp PowerPC does not have networking which has been a larger problem with the operating system; this is in part due to OS/2 x86's initial design around the 286 and reliance of 16-bit drivers that are hard-coded to the x86 instruction set. Porting them wouldn't have been an easy task and it looks like they ignored them for exactly that reason.”
However OS/2 Warp PowerPC did have a solid DOS emulator as well as support for x86 applications included as part of its apparently somewhat shaky Win-OS/2 compatibility. For more, I highly recommend checking out this article, I’ll put a link to it on the screen.
One more thing of interest. Some of the technology developed for OS/2 PowerPC had a significant amount of development in it, and even after this edition failed, IBM apparently back ported some of it to the main version of OS/2 Warp. I recommend reading this other article I’m linking to for more information on this.
OS/2 Warp 4
In September of 1996, the last major version of OS/2 was released into the rapidly shrinking OS/2 market. Probably due to Paramount’s reaction to the Star Trek internal codenames for previous OS/2 versions, this version of OS/2 was developed under the codename “Merlin”, but still retained the “Warp” consumer branding. When it was released, OS/2 Warp 4 was arguably the best all-around operating system to see wide release in the 1990s for consumer PCs.
And obviously I expect my comment section to light up for saying this, but look at what the competition was. By this point the Macintosh was a niche product, Copland was about to go down in flames, and the clone program was about to mostly cannibalize Macintosh market share instead of taking share away from Windows. And speaking of Windows, it was still on Windows 95, and when Windows 98 came out, it still wouldn’t match Warp 4’s features and stability. NeXTStep OS was still available, but few people were buying it and it was hardly in wide release. BeOS was obviously a very solid, stable operating system but it never got very far off the ground, few applications were written for it, and few people outside of techies had ever heard about it.
Linux in the 1990s may have been getting a lot of buzz in the tech world…but it was only in the tech world. And its legendary user unfriendliness was in full bloom at this time. And I say this as someone who up until recently had been running Pop!OS for the past several years, and doing all of the video editing and graphics creation for this channel on it. That all changed when I got tired of playing “what will break this time” when doing system updates, but that’s a different story. The point being that Slackware and Red Hat were hardly in danger of threatening Microsoft’s hegemony in the 1990s. Also I feel like I should mention that the team over at system76 have built a great distro in Pop!OS, and had I been running it on one of their Thelio systems instead of my custom built tower, I’m sure I wouldn’t have had the issues that recently forced me to grudgingly move my video editing rig back to Windows.
The only operating system that could have given Microsoft a run for its money by running on the computers that the vast majority of computer users owned, and was backed by a company with deep enough pockets to go toe to toe with Microsoft, and was a solid, stable operating system that also ran quite a few Windows programs seamlessly and had excellent DOS compatibility..was OS/2 Warp 4. For all that I still think that including Windows compatibility was a mistake.
Warp 4 was an excellent release that fixed, among other things, the Synchronous Input Queue bug that had been one of Warp 3’s more noticeable problem spots. Rather than the mildly confusing Red and Blue versions of Warp 3, Warp 4 instead had Windows 3.1 compatibility built into every version. Although I assume that this meant that IBM was back to paying Microsoft for every copy sold.
Warp 4 could also be installed over pretty much every previous OS/2 release, apart from OS/2 Warp Server 4, OS/2 for SMP, and weirdly enough Windows NT? I find it very odd that IBM’s official guide to Warp 4, called “Up and Running OS/2 Warp Version 4” felt the need to specify on page 17 that Warp 4 couldn’t be installed over NT. Was this a thing that people were trying to do? Early versions of NT could run OS/2 1.x applications, as according to a Byte article from January of 1994, that was one of five application modes that NT made available to its users. This was undoubtedly a hold over from when NT was originally being developed as OS/2 version 3.
Given the confusion that people tended to have with the whole “A Better DOS Than DOS and a Better Windows Than Windows” tagline, maybe the three confused NT users who had been running the umm…limited number of OS/2 1.x applications had for some reason assumed that they were actually running OS/2 instead of NT, and had thus attempted to upgrade from NT to Warp 3 and been upset that it didn’t work.
Warp 4 also added voice input, although I am mildly skeptical of how well it worked, given my own experiences with voice input using Apple’s PlainTalk technology back on the classic Mac OS when I was a kid, it was the sort of thing that was cool to show off but had very little practical application for daily use. However IBM’s software announcement of Warp 4, titled “IBM OS/2 Warp 4 – The Easiest Way to a Connected World” proudly boasted that you could “Navigate the Internet or your desktop with your voice, dictate responses to e-mail, and create speech macros to simplify repetitive tasks.” If anyone watching this had any experience with using this feature in OS/2, I’d love to hear from you in the comments.
But the war was already over, and OS/2 Warp 4 didn’t have a chance of digging its way out of the giant hole IBM buried it in.
Of course IBM was still IBM, and thus had to add an additional layer of incompetence on top of OS/2 Warp 4’s already impossible odds. Specifically in pricing, where IBM once again forgot that they were not the market leader, but were in fact attempting to successfully regain market share from Windows, tempt MS-DOS holdouts to make the switch, and solidify their position as a strong runner-up in the modern operating system wars. And so…they decided to price OS/2 Warp 4 at 249 dollars for the full version, or 149 dollars for an upgrade from a previous version of OS/2.
Now, this was significantly better pricing than early versions of OS/2 had enjoyed, but it was still higher than Windows 95’s launch price of 209.95, and of course previous versions of Windows were much cheaper on top of being easily pirated. Not to mention all of the OEM’s who got dirt cheap pricing on including Windows with their various PC compatibles. What IBM should have done is significantly undercut Windows’ pricing, push dirt cheap bundle deals to OEM’s hard, and go on an aggressive all-out marketing campaign to show that OS/2 Warp was back and better than ever. But that sort of clever positioning was anathema to IBM, and thus OS/2 launched with more handicaps than were, strictly speaking, necessary.
Plus of course its greatest handicap, which was the fact that IBM had basically told everybody not to consider buying it. Unsurprisingly, OS/2 4 Warp had no momentum, no buzz, and no success. IBM had finally succeeded in putting a stake through OS/2’s heart once and for all.
I will say though, that while it’s hard to know exactly what OS/2’s market share was at any given time in the 1990s, an interesting data point can be gleaned from the testimony IBM’s head of network computing and software services, John Soyring, gave in 1998. We have met him before when he introduced OS/2 2.1 on the Computer Chronicles. Here, he was the sixth witness that the federal government called in their ongoing antitrust lawsuit against Microsoft. According to Soyring’s testimony, OS/2 had roughly a six percent market share at the time. By way of comparison, on January 6, 1998, CNN had an article that said: “Apple's market share in the United States slips to 4.0 percent in the first quarter before recovering slightly to 4.4 percent by the third quarter.” In other words, even several years after the total implosion of OS/2, it still had about half again as many users as the flailing Macintosh did.
The final OS/2 release was in November 2001 with a fairly minor release called the OS/2 Convenience Pack 2, also known as the Merlin Convenience Package, which was built off of OS/2 4.52.
OS/2 was dead and gone. The OS/2 specific magazines and publications were long since defunct by this point, the remaining user base was rapidly shrinking, very little software development was going on apart from things like internal software development at companies that were still relying on internally developed OS/2 applications. OS/2 was a product stuck in the past, a relic of an era before wireless Internet was ubiquitous, back when Compuserve was still a separate thing from the internet. Windows had leaped ahead to Windows XP, switching to the NT codebase and continued world domination. OS/2’s support for Windows was still limited to Windows 3.1, and 16 bit applications. OS/2 was a fading operating system that was less and less relevant as the world moved into the era of smartphones and high speed internet.
As legendary PC columnist John Dvorak, himself an OS/2 user, said regarding OS/2’s downfall: "Microsoft created a lot of interest, a lot of pre-publicity; they did a great job of promoting the Windows product before it came out; and IBM was very casual about it, thinking that people were going to flock to them, the guys who built the better mousetrap. It was just poorly handled. IBM had gotten a lot of breaks earlier, before Presentation Manager. You read a lot of stuff about how important OS/2 was going to be, and then they just never followed up on it. There's a huge difference between the way the products were promoted and rolled out. I have nothing but sympathy for OS/2 users, being one myself, but on the other hand c'mon, it was so obvious that this was never going to go anywhere."
ArcaOS
And yet…somehow OS/2 Warp is still available for purchase today. Not from IBM, but rather from a third party company that licensed the rights to OS/2 and took over development in 2015. You heard correctly, OS/2 is still being actively developed by a company called Arca Noae. This version of OS/2 goes under the name of ArcaOS and is currently sitting at version 5, with an upcoming version, 5.1, codenamed Blue Lion, expected this year.
ArcaOS is based off of the final commercial release of OS/2 Warp 4, as previously mentioned this was released back in November of 2001. And ArcaOS has benefitted from some significant enhancements and improvements, with support for symmetrical multiprocessing, new drivers for modern machines, updated networking support, additional bug fixes and even an updated kernel.
ArcaOS is not free, rather it costs 129 dollars for the personal use or 229 dollars for a commercial license. It is aimed at users and businesses who still need or want to run OS/2 applications, and also allows for the running of 16 bit Windows or DOS applications.
And ArcaOS is not the only company that’s been involved with OS/2 since IBM abandoned it, back in 2001 two companies, Serenity Systems and Mensys BV jointly started developing and selling a new version of OS/2 Warp 4 called eComStation. There were issues with funding, the rights wound up held by a sister company now called PayGlobal Technologies BV and there hasn’t been a new release of EcomStation in well over a decade. It is however still available for purchase from eComStation.com, although the website is umm….barebones.
Wrap-Up and Final Thoughts
In closing, I think there is no better summary of OS/2 than the one from the final issue of OS/2 Professional, in February of 1996. This issue was a lot slimmer than previous ones, coming in at only fifty pages in total, when previous editions had reached a hundred. The rapid collapse of OS/2 had indeed engulfed it in a shockingly short amount of time, just six months after everything fell apart and before Warp 4 had even launched. By this point it was very clear to even the most ardent optimist that OS/2 was done for, dead but not yet buried.
In Edwin Black’s final Publisher’s Memo, titled Right Now, he said the following: “We will be remembered as pioneers who had the vision and the guts to put ourselves on the line in the face of insurmountable stupidity by IBM and insurmountable cunning by Microsoft. Yes. Someday all corporate Microsoft users will have what we have today—a superior OS…drag and drop, crash resistant connectivity, speed, resource-lean multithreaded multitasking and backward compatibility. Yes. Someday all of Microsoft’s SOHO and personal users will have these same benefits changing their lives and the way they connect to the world from a single powerful workstation. But we in the OS/2 Community had it first in 1992 and 1993, not at the turn of the next century.“
The memo goes on to give a great summary of the OS/2 user base, the fans who had so fervently supported an operating system that could have been a serious player, but was repeatedly undercut and inadvertently sabotaged by its own creator, IBM. “No matter how Big Blue fumbled and stumbled, no matter what the setback, no matter what the missed opportunity, you—the OS/2 loyalists—would not relinquish the prospect that another horizon would clear beyond the clouds. Please take a moment after reading this column to congratulate yourself for standing up for an ideal and forcing the rest of the world to do it right—if only by your example. It was you who wandered through the desert of marketing incompetence, broken promises, false prophecies and profound adversity to lead the way.”
I don’t really think I can add much more to this. OS/2 was a remarkable product that deserved a lot better than it got, and somehow saw a lot more success than one would expect from a product so relentlessly self-sabotaged.