Fall of OS/2 Extended Thoughts

Some random musings that didn't make their way into the full video

#1 Desktop Software Division 

The Desktop Software Division was actually the direct creation of Jim Cannavino, who created the division in either 1988 or 1989 (In Search of Stupidity says 1988, Big Blues: The Unmaking of IBM says 1989). Cannavino wanted it to operate as a small business, with the goal of not only developing its own applications but also finding good applications others had made, and then publishing them under IBM’s banner.

Cannavino gave the division over to a man named Fernand Sarrat, who was somewhat of an IBM sybarite. He apparently greatly enjoyed the perks that came with being an IBM executive, having an assistant to pick up his dry cleaning, get him sodas, and drive him to work. Although in fairness, Sarrat was also a smart guy, who from everything I can tell, took his job seriously. He also came from money himself, and claimed that he frequently didn’t even bother to submit his expenses to IBM for reimbursement. 

Sarrat did his best to get things moving, but had almost no luck. Software companies viewed IBM as having little to no ability to move software, and thus were fairly uninterested in having IBM publish their offerings. As mentioned in the video, the division did see its best success with DisplayWrite, a port of the mainframe version. Its other internal applications didn’t stand out, but from everything I can find out were competent, if unremarkable, offerings whose struggle to gain recognition was probably mostly due to IBM’s inability to market things correctly.

Additionally, the group was hampered by IBM’s massive bureaucracy, something that Cannavino should probably have done more to shield them from. Or did anything at all really, from what I can tell he didn’t seem to recognize the need to give the business support while also protecting it from IBM’s massively bloated corporate structure. 

The other huge problem the division had was in its incentive structure, or rather the lack thereof. The programmers working for it were IBM employees, under IBM’s standard pay structure. If they developed an OS/2 application that really took off, or even turned into a killer app, they stood to gain nothing but a bonus and maybe some limited internal recognition. This is a problem that occurs over and over again with big companies who attempt to set up an internal division that is supposed to act like a small, fast independent business but is hampered by being tied to a huge corporation. Still, this is not an unsolvable problem, and could have been greatly mitigated by enacting some form of royalties for a programmer or programming team who created an application that sold well.

In 1990 Sarrat came really close to a major coup for the division, and potentially a huge validation of their existence, when he spotted Borland’s Paradox database. Paradox wasn’t really selling well yet, Borland had bought the company that made it but was still figuring out what to do with it. Sarrat wanted to buy a major chunk of Borland, then use that interest to build a powerful suite of applications around Paradox. Details are sketchy, but to me it sounds like he wanted to create something that would probably have looked a lot like Microsoft Office. A powerful database, plus probably word processing and a spreadsheet. Sarrat got a tentative deal put together to buy ten percent of Borland for ten dollars a share.

Unfortunately the deal had to be run past multiple levels of the IBM hierarchy, all the way up to Akers, and the mainframe division managed to squash it on the grounds that Paradox didn’t fit into IBM’s mainframe business. No really, that was apparently the reason. Sarrat managed to get this decision overridden, but it took so long for him to do so that Paradox started to sell incredibly well, and Borland quite reasonably no longer wanted to sell at 10 dollars a share.  

So why do I fault IBM in the video for canceling them, since they seem to have produced very little? Well, freed from the constraints IBM smothered them with, they at least could have improved OS/2’s dismal first party software offerings at each version’s launch, and even in the worst case scenario they would still be able to provide something. And something is better than nothing, especially for a product like OS/2 that needed all of the help it could get. Additionally, with some form of profit sharing or royalty structure, even a modest one, they would have been far more incentivized to act like a small, fast moving company and may have produced not only something really really good, but possibly even a killer app for OS/2.

It frequently only takes one killer app to make a platform viable, whether that was VisiCalc for the Apple II, PageMaker for the Macintosh, Deluxe Paint for the Amiga or Halo for the Xbox. But OS/2 never got a killer app, and absent the marketplace penetration that would have at least vastly increased its chances of an outside developer creating one, the next best chance for one would be from a small internal team that acted like a startup. The Desktop Software Division was the closest thing IBM had to that, and it definitely had talented people working for it. Had IBM handled it properly, that division at least had the potential to come up with an OS/2 killer app, something the platform never had, and the lack of which greatly contributed to OS/2’s demise. 

#2 Inclusion of Windows compatibility

  I strongly disagree with IBM’s decision to add basically full (or close enough) Windows 16 bit compatibility to OS/2. DOS compatibility was one thing, it gave OS/2 access to a large body of existing software and I think was a helpful selling point in the short term. But including Windows compatibility really makes zero sense from any strategic standpoint. Tactically speaking, you can make the argument that it widened OS/2’s contemporary appeal, and enabled it to somewhat benefit from the vastly increased body of applications that flooded into the Windows market as it rose in power. However strategically it was a trojan horse that crippled IBM on multiple fronts.

First of all, as mentioned in the video it allowed application developers to avoid taking a side. They could simply develop for Windows, then use OS/2’s Windows compatibility to claim that their application was “OS/2 compatible.” This then had a knock on effect where more and more software was developed for Windows, which drove more and more sales of it, which meant that there were more and more incentives for developers to focus on Windows first, and then eventually Windows only. Especially in the early 90s, this proved a key asset to Microsoft as it broadened its market share.

Secondly, the inclusion of Windows meant that every copy of OS/2 sold with it…required IBM to send Microsoft royalties. I mention this in the video as well, but really think about how bad a decision this was. The royalty fee ensured that OS/2 cost more than Windows at retail(somewhat obviated with the later separate versions of OS/2 and OS/2 with Win/OS2 at the price of increased consumer confusion), which was basically the only place it directly competed with Windows since IBM almost entirely ignored the OEM market. The royalty fee also poured tens of millions of dollars into Microsoft's coffers, money that could be channeled directly into more sales of Windows and more incentives to OEMs to bundle Windows with their systems. It should go without saying that any plan that requires you to give your enemy a steady stream of revenue to attack you with…is a really, really, bad idea.

Thirdly, the inclusion of Windows 16 bit compatibility kicked a particular can down the road that would have to be dealt with at some point. Specifically, how long would IBM be forced to include this, once it set the expectation? When Windows 95 and win32 came out, IBM elected not to even try to include compatibility (by this point OS/2 was dying anyhow), and I highly doubt Microsoft would have been amenable to it anyhow. Even if IBM hadn’t totally kneecapped Warp and it had been able to carve out a solid niche for itself, failure to include win32 compatibility when the expectation was that OS/2 could run Windows programs…would have caused significant market confusion and given Microsoft an easy ad campaign to pound OS/2 with. Something like “why worry about whether a particular Windows program will run under your OS/2 install, just pick up a new install of Windows 95 (heavily discounted if you send in an OS/2 retail box) and you can not only access all of the new Windows95 applications, but all of your existing Windows applications will run seamlessly as well!”

A nice contrast with Apple’s Windows compatibility can be drawn (or at least its compatibility in the Intel era). Apple provided Boot Camp, but did not provide a Windows install or disk, the user was on their own to get that. If they needed to run Windows programs, it made it pretty easy to install Windows and just boot the Mac into it when needed, but it never confused users or forced Apple to pay royalties to Microsoft. Sure there was Parallels, which gave a similar experience to OS/2’s Windows compatibility insofar as it enabled Windows programs to be seamlessly launched from within OSX, but this was a third party program that helped Apple, without forcing them to commit to explicitly including Windows compatibility. 

At the end of the day, IBM’s inclusion of Windows compatibility did far more harm than good, and played a major role in preventing a healthy ecosystem of apps from growing up around OS/2. It's certainly not the only reason OS/2 never received any killer apps(see point 1), but it's certainly one of the major contributing factors. 

#3 Using same marketing agency as Microsoft 

Not going to spend a lot of time on this, but I find it rather hilarious that the marketing team of Ogilvy & Mather that IBM brought on in late 1994 after firing their previous marketing team…was apparently the same marketing agency that Microsoft was also using. At least InfoWorld in 1994 cites Ogilvy & Mather as having landed the Microsoft Corp account in January of 1994.

I’m not entirely sure what all they worked on for Microsoft, but the fact that they were involved with both Microsoft and IBM in the same year is definitely entertaining.

And Ogilvy & Mather is still working for IBM, at least as of 2022, so it appears that the OS/2 marketing fiasco didn’t damage the relationship. Justly so in my opinion, as IBM should bear the bulk of the blame for OS/2 Warp’s complete marketing failure.

#4 DeScribe 

Since I never mentioned DeScribe in the video, I am sure it will get brought up in the comments. And deservedly so, as DeScribe was one of the standout applications for OS/2 and one that deserved a far better fate than the one it received when OS/2 collapsed on top of it. 

DeScribe was a word processor for OS/2, for its day it was a powerful, and full featured application that quickly became the gold standard for word processing on OS/2. 

I may do a full video on it at some point, and I am not going to do more than hit the highlights here, but it was a very impressive program. It really got rolling when DeScribe 4.0 was released for OS/2 2.1, apparently receiving excellent reviews from even non-OS/2 specific publications. One ad for DeScribe 4.0 quotes PC Magazine as saying “It…performs amazing acrobatic feats of editing and display that leave rival programs gasping in envy” as well as PC Week lauding DeScribe 4 as an application that “sets the standard for OS/2 applications by taking excellent advantage of the operating system’s advanced features.” Basically DeScribe was seen as not only an excellent word processor, but also an excellent example of what an application coded specifically for OS/2 and its feature set could do. 

DeScribe was created by a start-up, and was solely targeted at the OS/2 market. And so of course when OS/2 imploded, this excellent application went down with it. Details are scarce, but it seems like DeScribe 5.0 was the final release (and it also was apparently finally released for Windows as well), and it looks like all development ceased in 1996 or 1997.

---