Too many people have speculated why OS/2 faded into the sunset while Windows reigned supreme. I want to explore some of the bigger theories and then offer up one of my own. The fact remains that OS/2 was the operating system that should and could but never would. I wanted OS/2 to work. I remember the Visual Basic class I took all the way back in the summer of 1994. I remember the instructor going around the class having us introduce ourselves. I remember saying how much I hate Windows (it was 3.1 at the time) because it was such a hacked and slashed together operating system, and that I really wished we would’ve gone the “OS/2 route”. In my mind, Windows 3.x was nothing more than a glorified version of DOS Shell, although I later learned that Windows was a real operating system; an operating system you add to pay twice for because you couldn’t run it by itself. All kidding aside, even in my late teenage years I stayed loyal to the OS/2 brand. I even ran it on my PC at home, albeit in dual boot mode because there was a lot of stuff that Windows 95 could run that OS/2 could not. I owned two versions of OS/2, “OS/2 for Windows” which was OS/2 version 2.1 without a standalone copy of Windows where you used your own copy. I also purchased OS/2 Warp 4 in 1996 when I was sick and tired of Windows 95 and the constant reboots.
The superior operating system is rarely the one that becomes the most popular. The one that usually wins, at least in the very beginning, is the cheapest. That was the case with MS-DOS, or IBM’s version PC-DOS, in the very beginning. They were able to undercut the closest competition by a price point of 6-1 ($40 for PC-DOS, $240 for CP/M). CP/M was a much more mature and superior version of the operating system that already had worldwide success. The power of the dollar ruled supreme here because most end users buy an IBM PC or compatible couldn’t really care less what the differences were between the two. With the runaway success of Windows 3.0 and the lackluster sales performance of OS/2, Microsoft decided that OS/2 didn’t fit their long-term strategy. It’s hard to pin down the exact date of the breakup, but it occurred somewhere between 1990 and 1991. Microsoft pushed on with the next wave of Windows products while IBM concentrated solely on the OS/2 line. The breakup still leaves open wounds to this day as I don’t think Microsoft and IBM have ever decided to kiss and make up. As a matter of fact, IBM pushes Linux as their official platform of choice today, although they do offer Windows to customers who want it.
Theory #1: Microsoft OEM Bullying
This theory suggests that Microsoft used their position of domination in the marketplace to bully PC manufacturers and to force them to commit to Windows. I don’t deny that Microsoft has used this tactic in the past with CP/M and the like, but if customers demanded OS/2, the OEMs would have conceded. Besides, most OEMs offered OS/2 as an alternative operating system. Again, I believe economics played into it, and the relatively low cost for Windows and DOS (pre-Windows 95) and Windows 95 afterwards was the main consideration here. Not much, if any, Microsoft bullying was needed. There is a reason OS/2 had a higher price tag. Quite simply, it was a lot more operating system for your money. The problem was that most of your basic users just needed the bare minimum configuration. Windows offered that. It was hard to explain to your average Joe computer user why they needed the High Performance File System.
Theory #2: OS/2 Was “Too” Compatible
This is my favorite theory, just because it sounds so outlandish, not because I think the theory is plausible. This theory suggests that the reason OS/2 failed is that they truly lived up to their claim of a “Better DOS than DOS and a Better Windows than Windows”. Simply put, there was no incentive for software developers to write OS/2 programs since OS/2 would still run the same DOS and Windows programs that were being developed. Just write a good Windows-based program and it will run on OS/2 and Windows. You can target all the customers instead of having to develop several independent product lines. The reason I can’t buy this theory is that, if true, what would be anyone’s incentive to write 32-bit Windows software supported by Windows 95 and NT? Why not continue to develop 16-bit software all around and guarantee that everyone can run it regardless of OS? The answer is that you want to develop 32-bit software because it is theoretically more stable, and you have a lot more memory to work with. There are other reasons, but we’ll stop there for simplicity’s sake. Let’s just say that a 32-bit program can be more feature rich than a 16-bit program. If the demand was there for OS/2, I guarantee that software vendors everywhere would’ve been jumping on the bandwagon. IBM really didn’t help their case by charging so much for their software development kits. Good, low price development kits and some conversion assistance software may have gone a long way, but, again, was not the real reason for OS/2’s demise.
Theory #3: IBM Could Not Market
There is no denying that IBM as a whole did not fully support OS/2. Some at IBM even saw OS/2 as an embarrassment. If you go to YouTube and look at some of the OS/2 commercials, it almost appears like IBM didn’t even try. Some of the commercials didn’t even show the product in action, it just talked about it. There was a book made some years back called “IBM Way: Insights into the World’s Most Successful Marketing Organization” (click this link for the book). Granted, the author, Buck Rodgers, was a former employee of IBM. I thought it was a joke when I first saw the name. But, there is a case made of why IBM is a very successful marketing company. This is actual proof that IBM can market. Although, one could make a case that IBM was bad a directly marketing to consumers. Again, not entirely true since the IBM PC was a successful consumer product. IBM made several attempts to really push OS/2 but they all came up a little short. Their biggest push was in 1994 with OS/2 Warp 3.0, trying to capitalize on the delay of Windows 95, or “Chicago” as it was known at the time. The consumer market just wasn’t ready.
Theory #4: OS/2 Began as a 16-bit OS
There is no doubt that 16-bit programs, even if it runs in protected mode, are crippled in comparison to their 32-bit 386 counterparts. But, would that spell doom for OS/2? I really doubt it. Let’s look at the time tables. The Intel 80386 was released in 1985. Since OS/2 was released in 1987, relatively few people could actually run it. In 1987, there were quite a few computers still released with 8088 chips. The 386 wouldn’t go on to be a dominant chip in new PCs for at least another 4-5 years after inception. In reality, IBM/Microsoft had no other choice than to create OS/2 as a 16-bit operating system. If they would’ve waited until the 386 was a dominant player, there’s a good chance that someone else would’ve stepped in. OS/2 1.x was not a success, and, in most ways, probably a failure. However, I don’t happen to believe that starting out as 16-bit doomed OS/2. If it did, why didn’t Microsoft Windows, which gained most of its success after the 16-bit release 3.0, suffer the same fate? Looking back, IBM had to launch as 16-bit. There was simply no other choice at the time. When it was first released, people balked at the 2MB+ RAM requirement. I couldn’t imagine what the outcry would be if the requirements started out being 4-6MB instead.
Grantster.com Theory: IBM – Too Big to Succeed
IBM’s failure goes all the way back to the original PC. In an effort to enter the entry-level home computer market, IBM decided to develop a machine using off-the-shelf components. Really the only thing that was proprietary to the IBM PC was their BIOS. IBM actually published the source code for their BIOS. Their BIOS was the only thing that made their system “proprietary”. IBM wasn’t worried since they knew that if anyone tried to use their BIOS, they could sue them into submission. They didn’t take into account that, since the BIOS was published, that someone could reverse engineer it because they knew exactly what is was supposed to do. The inadvertent result of this was that quite a few clone manufacturers sprung up almost overnight, the largest being Compaq. This resulted in very stiff competition with IBM. Spring forward from 1980 to 1987 when OS/2 was released. Originally, the arrangement between Microsoft and IBM for OS/2 was going to be identical to their arrangement for MS-DOS. As a refresher for the uninitiated, Microsoft licensed MS-DOS to IBM to be sold as PC-DOS. Microsoft then had the freedom to license MS-DOS to anyone else who wanted to use it and to collect royalties.
This arrangement would’ve worked if Microsoft chose to continue their partnership with IBM. However, when Microsoft “divorced” IBM, it was up to IBM to license their operating system directly to OEMs and end users. Licensing to end users was no issue; you just sell it to them and they buy it. But most of the time, the operating system was purchased along with the computer when you bought it, which is the OEM model. Well, most OEMs were direct competitors with IBM. Although the OEMs may not have completely trusted Microsoft, they were not direct competition with them either. IBM has the unfortunate task of trying to court the very people they were trying to run out of business. How do you do this? The answer is, you don’t. Or, at least, you don’t do it very well. And IBM didn’t. That’s not to mean that they didn’t try. IBM also had an issue with the availability of drivers. Most hardware manufacturers write drivers for their own equipment. Many OEMs had no desire to write device drivers for their competition. There is no way that even the biggest companies could overcome these odds.
So, the question is, how could IBM have succeeded? I can only think of one way, and I’m still not sure it could’ve worked. IBM would have to create a brand new company, not a subsidiary of IBM but an actual new company with completely different management. That company would have to treat IBM as any other customer and show no favoritism. That alone wouldn’t work. They would also need to show all OEMs that they are willing to work with them in any and every way possible. They would have to beat Windows in price and give extra incentives to the OEMs to convince them. They would have to give away development kits.
Conclusion
I was, and still am, a fan of OS/2. I wanted OS/2 to work and succeed. The fact that there was no way that IBM could compete in the industry was their downfall. OS/2 is still sold today from Serenity Systems and it is called eComStation. It has better support for OS/2 on modern hardware, although some of the non-BIOS hardware will not work. I really want a copy of it; I just haven’t found an excuse to buy it just yet. There is probably no practical value in it, just something to play with. It’s too bad that very few people ever got to experience it.
Grantster.com - The WORST In Computer History - Returns! 
The book you want is Paul Carroll’s “Big Blues: The Unmaking of IBM”
It covers the OS/2 story in some detail, including the “breakup” of
IBM and Microsoft with regard to OS/2 development. OS/2 was
essentially doomed once Windows 3.0 became a success in 1990.
Coming from a commercial background and supporting customers with kit running CP/M, DOS, UNIX, Novell, OS/2 & Windows I suppose I’ve a different viewpoint.
OS/2 was doomed because it was almost impossible to offer a standardised support model. The seemingly infinite variety of configuration options meant that end users could (and in my professional experience) did, all too frequently ‘play’ with the display settings to the point of crippling their systems.
Over the course of two years I attended only 1 genuine OS/2 fault out of probably 300 client visits, circa 200 were hardware issues and half that number again were simply to restore default / as shipped settings. These calls were invoiced back to the customer because they were avoidable resulting in massively disproportionate support costs.
With the networked release WFWG and Windows 3.1 the speed advantage of OS/2 was lost, and combined with the ability to lockdown those OS releases against user adjustments the demise of OS/2 in the rapidly expanding office environment was assured.