OS/2 Past, Present, Future

(part 2)
Bradley Wardell
Stardock Systems, Inc.

December 1997


Copyright notice
This material is protected under international copyright. If you wish to reprint this article in part or in full, please contact bwardell@stardock.com for permission. You may also visit http://www.stardock.com to see this article and related articles.

I work for an OS/2 software company called Stardock Systems, Inc. We develop and publish
32bit OS/2 software for both corporate OS/2 sites as well as individual OS/2 users. Our
products range from complete desktop environments for securing and standardizing corporate
desktops (Object Desktop Professional), to utilities to allow corporations and "power users" to
recover from system hangs (Process Commander) to consumer entertainment products such as
Galactic Civilizations, Trials of Battle, and most recently Links OS/2.

PSP stands for Personal Systems Products, a division of IBM..
Chicago stands for the codename given to Windows 95.
All these articles involve the client side of OS/2.

Part II: OS/2 Present

A final look back
After the first article became widely read I received a lot of email from people who had lived through the OS/2 2.0 to OS/2 3.0 days.  I thought I’d add a little more about the past to give further background into this article.

The email I received confirmed the rumor that OS/2 2.11 SMP was largely the work of 1 person who later left IBM.  Once that person left and IBM had gotten rid of most of the contractors at Boca Raton Fl (IBM decided to centralized PSP in Austin) the brain drain was so much that OS/2 would never get a client SMP version again.  OS/2’s kernal on a P-66 took 2 hours to compile.  OS/2 took about 40 minutes to compile on a 2 processor OS/2 2.11 SMP machine (4 P-66’s).  IBM would not allow OS/2 internal builds to use the SMP compile.  If each compile ate up over 2 hours, imagine how much better OS/2 Warp could have been if the engineers had had an extra hour and 80 minutes after each build to spend testing, fixing, adding, etc.?  The fact that OS/2 had an SMP version so early could have made a huge difference the high end market.  Unfortunately, IBM failed to capitalize on this and eventually Windows NT (which still can’t match OS/2 2.11’s SMP scalability) took over as the primary SMP client.  By the time IBM got around to showing how cool OS/2 SMP was, it was too late.

Background on OS/2 Warp 4
A lot of the email I received focused on IBM’s inability to market OS/2.  IBM did try very hard to market OS/2, they just couldn’t figure out how to effectively market it to all  their market segments at once.  IBM spent the bucks, it just didn’t spend them effectively.

IBM also treated OS/2 ISVs very well.  Stardock benefited greatly from IBM’s encouragement and helped our growing little company learn the ropes of doing pre-loads, site licensing, demonstrating, and yes, marketing.  IBM UK and IBM Germany did a heck of a lot to help Stardock especially.  People ask why we stick with OS/2 and one of the reasons is that IBM’s terrific treatment of us during this time period earned our long term loyalty.  Unfortunately, the people who made all this happen at IBM are mostly gone now.

At a particular IBM meeting in the Fall of 1995, one of the executives at O&M (IBM’s advertising agency) presented IBM’s next stage marketing campaign (The infamous Nuns ads and other solutions for a small planet ads).  Once he asked if there were any questions I quickly raised my hand…

Now before I tell you what my question was, I want to make clear that up to this point I had never “burned” any bridges with anyone at IBM.  My question though quickly set flame to one of my bridges with PSP…

In front of some of IBM’s top customers, I asked “Those Warp 3 ads with the surfer people…what exactly were you thinking?”  The O&M guy responded “We wanted to show that OS/2 was hip and cool to use.” I said, “Well, who exactly are you targeting? Are you trying to imply to the 15 year old market segment that if they use ‘Warp’ that they’ll get beat up less at school?  You could have shown how OS/2 was a good and beautiful OS that increases productivity but instead wasted the advertising campaign promoting OS/2 as something you use after surfing!”  At that point, the IBM customers laughed and then applauded for they felt the same way.   I knew my question would stir trouble but after seeing all the posts and meeting all the OS/2 users who were frustrated at how poorly the USA Warp 3 marketing campaign was launched, how could I not seize the opportunity to bring this up to the specific people responsible for those horrible ads?  It turned out that the presenter was one of the actual people who came up with them.  Ironically, my question helped cement the Stardock/Vobis pre-load agreement since they felt the way I did.  It should be noted that we are talking about the USA OS/2 ads.  The European ads were quite good.

Comdex 95, the end of the world.
It was by Comdex 95 that Lou Gerstner had told PSP that OS/2 Warp for the PowerPC had to be done.  IBM PSP failed.  IBM had invited Stardock to demonstrate in the PowerPC tent that year but we declined since we knew of this rumor and knew that Power OS/2 (as I liked to call it) would no way be done by then.

One of the things that made OS/2 relatively successful was that within IBM there were contingents of OS/2 advocates.  People who lived, breathed, and slept OS/2.  I won’t name names here but in Austin, in the United Kingdom, in Germany, in Denmark, and speckled elsewhere were people who made things happen for OS/2 even though they didn’t have executive level positions.  Because of them, they were able to get things to happen that would otherwise not have.  It was, for example, an IBM contractor in Boca Raton who “discovered” Stardock, got IBM to port DOOM (well before Microsoft had thought of it), got video tools put into OS/2, etc.  Most of IBM is pretty indifferent to the world.  Many IBMers think of their jobs as the thing they do between weekends.  So when you get the IBMer that is an advocate of something (like OS/2) he or she can have great impact.

I mention this because at Comdex 95 two things became clear.  #1 Because Power OS/2 failed to meet its deadline, PSP was doomed from a budget point of view.  And #2 Microsoft, ever paranoid of PSP, had spared no expense in promoting Windows 95 (and hinting that NT 4 would solve any remaining problems Win95 didn’t solve).  As a result, many of these advocates saw the light at the end of the tunnel as the freight train of the Microsoft marketing machine.  These IBMer OS/2 advocates had hoped to make OS/2 the dominate platform. But Comdex 95 made it clear that OS/2 would be, at best, a niche.  Thus began the great exodus of IBMer OS/2 advocates to either different parts (i.e. NON-PSP) of IBM or out of IBM entirely.  Even David Barnes eventually went to Lotus (even though he still loves OS/2).

Taking Stock of the Situation…
It’s January 1996 and IBM has to assess the situation.  PSP is to be trimmed drastically, largely due to its handling of PowerPC OS/2.  The Warp 3 campaign was expensive and many of the “13 million” OS/2 users left OS/2 as quickly as they came (I could write an entire article on how ridiculous the claim is that there are ‘13+ million OS/2 users’ is). The IBMer OS/2 advocates were largely gone. And to add insult to injury many of IBM’s OS/2 “Business partners” not only didn’t produce anything but blatantly ripped IBM off.

The long time OS/2 ISVs, which, unlike the DOS/Windows ISVs who had received millions of dollars to bring their products to OS/2, had not gotten any money and subsequently had not noticeably grown (except mainly for Stardock which had moved into the lucrative desktop management and utility market).  In short, IBM spent millions on DOS/Windows ISVs in an effort to get them over on OS/2 and these ISVs largely took IBM to the cleaners.  But IBM largely ignored OS/2-specific ISVs and thus they remained small <10 person shops in general.  Some people always give the suggestion that IBM should pay Windows ISVs to port to OS/2 – IBM tried that and got the short end of the stick.

But the fatal flaw (in hindsight) in the strategy is that IBM could have spent the money to help the OS/2 ISVs become big companies that would have been loyal and been able to produce “Class A” products instead of paying big companies to port their class A products to OS/2.

Again, even though to the average OS/2 user it may not seem like a big deal, OS/2 for the PowerPC is what did the most damage to OS/2 --in terms of wasted developer energy and in making the rest of IBM lose faith in IBM PSP (which in turn vastly cut PSP’s budget).  It wasn’t Microsoft that hurt OS/2 the most, it was OS/2 for the PowerPC.

So IBM’s options in January of 1996 were pretty limited.  They had blown much of their third party support money trying to get various (unnamed here) Windows ISVs onto OS/2.  Those companies largely took the money and never delivered anything. Since IBM was slow to enforce their agreements (and the agreements were structured in a half now, half later payment schedule) and now (1997) PSP doesn’t have the money to pay the other half so they just terminated all the outstanding agreements.

So now they didn’t have any money to help mature the existing OS/2 ISVs.  Most of the people who had worked on the core parts of OS/2 were contractors who were long gone (back in Boca Raton) so Warp 4 couldn’t have too many fundamental changes to it.  The marketing campaign would be very small due to budget cuts.  What were they going to do?

In February of 1996 I flew to Austin at the request of PSP.  At this point, Stardock’s market share in the OS/2 market made us the clear leader in independent native OS/2 software.  Since we were continuing to release more and more OS/2 products, IBM was interested in what Stardock thought about OS/2 Warp 4.  At this point, JAVA wasn’t even on IBM’s radar screen (yes, I realize that seems hard to believe but it wasn’t until Spring that JAVA really came into the scene).  Warp 4 was going to add in OpenDoc, Coaches (which is where a big chunk of development money went), an updated UI, Smartcenter (later called WarpCenter), a system object similar to Windows 95’s, and a number of other features that had been hanging around PSP and other divisions of IBM but hadn’t made it in before because of IBM’s rigorous testing methodology (which any IBMer could write books about).

At that meeting it became pretty clear that the development resources at IBM had disintegrated since the glory days of OS/2 2.11 and 3.0.  Basically, PSP no longer had enough developers who were familiar with the core code of OS/2 (much of OS/2 is still written in assembly).  This meant no dynamic caching, no multiple message queues (though one heroic PSP developer did much to work around the problem which is in Warp 4 and FP 17 or later), and no major changes to SOM and such.  When asked, with these issues in mind, what I thought many users would like to see in Warp 4 I said (in this order): Make WinOS2 look like OS/2. Modify the file dialog to see long file names on FAT.  Add TrueType support. Simplify the desktop (i.e. fewer objects on the desktop, sort programs by their type – applications, utilities, etc. instead of Windows apps, OS/2 apps, DOS apps, etc.). And make the networking more intuitive.  While some of the suggestions went in (true type, simpler desktop, and more intuitive networking) the biggest suggestions were left out.  Stardock had planned on putting in an enhanced file dialog into Object Desktop 1.5 but chose not to in hopes IBM would do it.  The same is true for making Windows programs look like OS/2 programs.  These two things alone would have made OS/2 much nicer to work in.  The True Type support was done very poorly, giving OS/2 fonts a very ugly look.

JAVA? Cool, who needs OS/2 ISVs?
Now, sometime in the Spring of 1996, a new mood began to perpetuate about the upper levels of IBM.  That mood was, “we’re going to JAVA, we don’t really need our existing OS/2 ISVs so we’ll not worry about what we do that might hurt them.”

Examples of this included screen shots of “Merlin” from IBM which were nothing more than Warp 3 with a modified PMMerge.DLL, Object Desktop installed, and NPS installed.  This created the impression that OS/2 Warp 4 (Merlin) would make Object Desktop obsolete. This devastated our sales at the time – a time when Stardock was co-sponsoring the IBM technical interchange with IBM and Lotus.  But it wasn’t just Stardock that IBM stepped on.  IBM implied strongly that Merlin would come with a full blown tape backup package which hurt ISVs like CDS and MSR.  And, IBM went on to imply that Merlin would come with security features which likely had effects on Pinnacle (the leader in high security OS/2 software) as well as, again, Stardock which was previewing Object Desktop Professional which secures OS/2 desktops.  So until Merlin actually shipped, we had to battle the erroneous perception that Merlin would have many features that competed with Object Desktop.

IBM also decided to publicly tell users that OS/2 should not be used for playing games.  This statement, only a few months after IBM aggressively lobbied ISVs to bring games to OS/2, had devastating effects.  For example, when IBM positioned OS/2 as a game platform, people quit their jobs to start companies dedicated to writing for OS/2.  Excellent games such as Trials of Battle, from Shadowsoft, Bug Eyed Monsters from Grinning Lizard, and others sold less than 1/40th of what Galactic Civilizations II had sold.  Sales from these games wouldn’t even pay the salary of a single developer to live on leaving some of these people with not just disappointment but serious debts!  IBM had nothing to gain by telling people what OS/2 wasn’t but they did so anyway.

So by summer of 1996, it seemed like aliens had taken control of once benevolent powers at PSP.  At this point, the well respected columnist Wil Zachmann began to notice the aforementioned budget cuts PSP had suffered (whether he realized this was due primarily to the PowerPC debacle and not just Microsoft I don’t know).  Unfortunately, he concluded that these budget cuts meant IBM was essentially abandoning OS/2 and publicly announced this.  IBM hadn’t abandoned OS/2, they had merely given up on trying to take over the entire OS market.  Make no mistake, IBM had wanted to make OS/2 THE general purpose PC OS.  But by mid 1996, IBM was positioning OS/2 as a niche OS.  They hadn’t yet figured out where it would fit but JAVA was looking promising and they had the voice type stuff floating around that PSP had managed to forcibly obtain from another IBM division.

Internal debating at IBM was between two warring parties: Either [A] Position OS/2 as a high end OS where people would talk to their OS and do poweruser type stuff with it or [B] Make OS/2 the ultimate JAVA client.  They were mutually exclusive goals but it wasn’t until after Warp 4’s release that IBM decided on a course (choice B).

During the summer, IBM vigorously courted Netscape to bring their browser to OS/2.  IBM, having learned from previous experiences with “business partners” assigned some of their best developers to work on sight at Netscape to bring Navigator to OS/2.

The Warp 4 Launch…
If anyone doubted that IBM had pretty much given up on its native OS/2 ISVs, a trip to the Warp 4 launch would bring them around.  Stardock and Pinnacle were the only long time OS/2 ISVs that had a booth at the launch.  Ironically, IBM didn’t even seem concerned that Stardock had actually released a product that integrated the user’s desktop into the Internet – Object Desktop Professional’s object advisors (patent pending).  Stardock was there because of our workplace shell leadership and Pinnacle was there because of their leadership in the security market.

IBM had never resolved what OS/2 Warp 4 was supposed to be for.  It was the JAVA client you could talk to that had a kind of working Win32 API (open32).  Problem was, JAVA on OS/2 was slow and implemented so poorly on base Warp 4 to be useless.  Voicetype was a niche that required a ton of memory and was incompatible with many programs that used sound.  And Open32…well, it had more issues than this article room for.

That’s not to say that Warp 4 wasn’t a good OS and a worthy upgrade – it is.  But a successful product requires it to have a defined market, good distribution, and be technically sound.  Unfortunately, most OS/2 users even now aren’t sure who OS/2 Warp 4 is for.

Sales of OS/2 Warp 4, to put it mildly, did not live up to the glory days of Warp 3.  Many people just didn’t know why they needed to upgrade.  OS/2 Warp 3 with FP26 is pretty incredible.  If Warp 4 simply brought a slower version of JAVA and VoiceType why should people (especially corporations) upgrade was the general consensus.  Many of the sales came from an unexpected source – end users who, as this point, IBM wanted to go away.

So as 1997 began, OS/2 Warp 4’s market share became clearly a niche.  And it became clear that regardless of what IBM said or did, OS/2 would remain the choice of many “power users”, consultants, and corporations.  It also became clear that, for the first time, OS/2 was getting a life of its own.   That is to say, IBM had provided this wonderful 32bit, multithreaded, objected oriented OS and third parties and users had gotten together and begun to actually take the reigns away from IBM.  “The OS/2 Marketplace” developer’s conference was created by the Phoenix OS/2 Society – not IBM.  In October 1997, an OS/2 trade show, created by OS/2 users called Warpstock arrived.  In short, it could be that 1997 was the year that the third party proponents of OS/2 stopped riding IBM’s coat tails and took action on their own behalf.  Actions that will expand its customer base in areas IBM wouldn’t expect, enhance the OS in ways IBM cannot, and define what market segments OS/2 is for since IBM doesn’t seem to be sure where OS/2 fits in.

This is an exciting time for OS/2 users and the OS/2 market.  IBM created an open architecture for OS/2 and has even retreated out of all the market segments except for the high end client/server markets allowing others to step in and fill the void.  How they may fill that void we’ll talk about in the next part…

OS/2’s future…(Next Month)