Alas, Poor OS/2; I Knew it, Horatio

By: Dennis E. Powell
Saturday, May 20, 2000 11:53:38 AM EST
URL: http://www.linuxplanet.com/linuxplanet/opinions/1864/1/

I Read the News Today, Oh Boy

The news last week was scarcely a surprise, but it had that sudden deflating effect that one gets when an ancient and beloved friend and teacher, terminally ill, finally passes on: IBM is pulling the plug on OS/2. No service is planned, but the mourners will comprise thousands, maybe millions of Linux users who first tasted independent thinking in computer software when they endured the horrendous installation procedure of what was otherwise the most powerful and magnificent of dosrivative operating systems, present company included. Having earned their stripes by just getting the thing on the hard drive, they learned that power, reliable multitasking, yes, a better DOS than DOS and a better Windows than Windows were theirs.

There will be no funeral, but an obituary is in order anyway.

From the days when it was still a joint production involving IBM and its pet software company, Microsoft, OS/2 was a sickly child, not expected by many to live long at all. The first version, devoid of GUI or native applications and capable of running but a single DOS app at a time, was released in December 1987. It had no mouse support, wouldn't support hard drives bigger than 32 megs, and was functionally crippled by IBM's insistence that it run on 80286 processors. A beta had been sold the previous summer to anyone willing to pay $3,000 for it.

IBM promised a graphical front end for the thing, and a mere 11 months later, on Halloween, did so in the form of a ghastly thing called Presentation Manager. (Microsoft was still on the boat; the second version of Windows was named Windows Presentation Manager 2.0.) Nevertheless, the computer press went wild. This, they said, was the wave of the future. Article after article about this great new OS appeared, and when Lotus released 1-2-3G for Presentation Manager, the ink-stained crowd went wild. Microsoft released applications as "family mode," meaning that if they were installed under OS/2, they were native OS/2 apps. (This was true, in fact, through Word 5.5, released long after Microsoft had declared war on IBM. The chief change in Word 6.0 was elimination of this feature. I'm talking DOS apps; this was a few months before Winword inexplicably replaced its version 2.0 with 6.0.)

There was OS/2 1.1 and OS/2 1.1 Extended Edition, which was like 1.1 but now actually good for something. Microsoft wrote OS/2 1.2, which by all accounts made DOS 4.0, that beehive of buggy releases, look like high art. IBM rushed out OS/2 1.3, which was highly successful well into the mid 1990s as the imbedded system in the majority of the world's automatic-teller machines.

Dry facts. But then there was The Movement.

Team OS/2

Almost from the beginning there were people in IBM, renegades responsible ultimately for things like the abandonment of the white-shirt requirement, who saw the potential in what they thought was a tragically mismarketed operating system. They were hackers and they knew that not all hackers worked at IBM. They would write little apps for OS/2: Text editors, programming tools. Somewhere around here I have a floppy that contains an employee-written program that allows OS/2 to run CP/M applications. They lobbied heavily for OS/2, even though most of them had nothing at all to do with the product. They called themselves Team OS/2. And while they could not spread the word outside of IBM, they could say that something big was coming. A new version? Could be. Who knows?

By the middle of 1992, Team OS/2's guerilla war was all but public. Semi-official OS/2 representatives appeared on CompuServe and Prodigy, where the great Mel Hallerman held court, and little fora were set up to discuss the upcoming OS/2 2.0, a 32-bit thing that would actually run not just OS/2 apps--of which there still were few and there would be a lot real soon now--but a whole bunch of DOS apps at once and, yes, even Windows apps! By now, Microsoft was engaged in an acrimonious divorce from IBM, but IBM had a prenuptual agreement and therefore had custody of the code, including Windows 3.0. Big Blue recompiled Windows with the Phar Lap compiler, resulting in smaller and faster code than the homebrew tools Microsoft had used for standalone Windows 3.0. And it was included in OS/2. Steve Ballmer, then just a Microsoft hitman as opposed to the Microsoft hitman, his current position, said he'd eat a floppy disk if OS/2 were released by the end of 1991. IBM obliged, releasing OS/2 2.0 LA in November. (LA meant "Limited Availability," which is to say beta, and Ballmer didn't eat a floppy, or if he did it went unrecorded.)

On March 31, 1992, OS/2 2.0 was released. A lot of people went to software stores to buy it. It wasn't there. IBM had made no plans, had no idea really, that actual people would want to buy it. (I went through the Westchester County, New York, phone book. IBM is headquartered in Westchester County, and the company's phone numbers make up more than half a page of the phone book. I dialed every number until, two-thirds of the way down, I found someone who had a shrinkwrap copy. I drove there and paid retail, in cash, no receipt, in a transaction that more resembled what I'm told drug deals are like than a software purchase. It was a copy that had been sent to an office that didn't know what it was or what to do with it; my guess is that they had drinks after work with the money.) Those who did manage to find it were startled to learn that it had no manual. None. All the box contained was a little installation pamphlet and some peel-and-stick OS/2 stickers, and a box containing a couple dozen 3.5-inch floppies. The first thing most new OS/2 users did was change the ribbon cable on their floppy drives, so that the 3.5, not the 5.25, was now A:.

It took some getting used to. You could actually intermingle it with your existing DOS installation, which was so paralyzingly weird that anybody who did it (I plead guilty!) quickly tried to undo it, and failed. The dual boot was not a good idea. Besides, OS/2 offered the High-Performance File System, which made better and faster use of hard drives. This involved repartitioning, reformatting, lots of stuff that has disappeared in memory in the way that traumatic events sometimes do. What's more, you needed to set up each DOS application to run correctly--give Procomm Plus access to the modem, for instance--producing what were called "extended attributes." For DOS hands, it was like landing on Mars and learning the ways of the Martians.

But then one got the hang of it, and my goodness was it good! Team OS/2 at first extended invitations to a few mere users chosen on merit, then opened up membership to one and all, and for a time the organization was even supported to some extent by IBM. There were user groups. Lee Reiswig, the legendary Blue Ninja and head of IBM Personal Software Projects, would even drop by user group meetings from time to time and talk about the wonders to come--"If you don't have a CD reader, get one," he said with a wink before the release of OS/2 3.0 in 1994. Users would organize presentations at software stores and meetings of unenlightened user groups: hordes of users doing volunteer work in support of a multi-billion-dollar company that had a great product and didn't know how to sell it.

Few Applications and No Press

The applications never appeared in any great numbers--if OS/2 would run Windows apps, why write to two APIs when you can write to one?--though the single best application that I have ever used on any platform was the OS/2-native DeScribe word processor. (DeScribe was a microcosm of IBM and OS/2, a great product insanely mishandled. At one point the company's president, James P. Lennane, announced that the next version would work only if a key disk, sent out semi-annually, were applied; then he withdrew that plan and said that if 1,000 copies were not sold in the first month he'd close the company, which did not inspire buyer confidence for some reason. Finally he backed away from that, too. He appeared on "60 Minutes" and said that his software was being pirated. And he ran for president of the United States. I am not making this up; he was in the New Hampshire primary, but I don't know if anybody voted for him. No surprise, he was an ex-IBMer.)

Led by Reiswig, IBM actually played for the consumer market with OS/2 3.0, which wasn't quite as hard to install and which contained a full (though pretty miserable) Internet suite and IBM Works, an OS/2-native productivity suite that resembled Microsoft Works in a number of ways. And for a few months (during which time the sales charts were mysteriously absent from PC Magazine), OS/2 3.0 was the top-selling software product in the country. Nobody noticed -- The New York Times, for instance, ran a three-inch wire story about the OS/2 3.0 rollout in the fall of 1994 on an inside page of the business section; understandable, in that the rollout took place at the Richard Rodgers Theater on Broadway, an entire three blocks from the office of The Times. This was in stark contrast to the orgy of coverage The Times did the following year when, on July 31, 1995, more than a month before its release, the newspaper devoted page after page to Windows 95 in an orgy of speculation that made it seem as if the chief benefit of Creation was that it made Windows 95 possible. In that section The Times finally did a piece on OS/2: IBM was discontinuing it, the story said. Finally, last week, The Times was relieved of its obligation to publish a retraction, something toward which it had apparently decided to take a wait-and-see attitude.

IBM actually advertised OS/2 3.0, sort of. The only thing anybody remembers was a television ad featuring elderly Belgian nuns croaking something in Flemish that the subtitles told us was a discussion of the joys of being on the Internet with OS/2. It seemed never to have occurred to anyone in a position of power that introducing this product to a new market might have included, in English for English-speaking countries, what the product was, what it did, and why anyone would have wanted it. The advertisements soon disappeared, but OS/2 didn't.

There was, later, OS/2 4.0, which dropped the notion of marketing to consumers (a batch of games, home accounting programs, and so on, called the "Family Fun Pak," was said to have been produced, but I've never talked to anyone who actually had the thing). Instead of IBM Works, there was now speech recognition--even a little headset. You would have to train the program to understand what you were saying, but then you could speak into the microphone and see the words appear on the screen, ready to be pasted into any application. (OS/2's clipboard was incredible. If you could get it onto the screen, you could cut and paste it--text, graphics, anything.) IBM had been selling the speech recognition stuff only a few months before as a standalone application for $1,000. Now it came with an OS that you could buy for a hundred bucks. Too bad they never much told anybody about it.

The Sentence is Handed Down

OS/2 had run up against a lot of obstacles, not the least of which was the Microsoft preload agreement structure that required original equipment manufacturers to pay for DOS and Windows for every processor shipped--this took a lot of the incentive out of paying for and shipping another OS, too. It was part of what led the Department of Justice's Antitrust Division to investigate Microsoft in the early 1990s. The DoJ folded and agreed to a settlement so flimsy that U.S. District Court Judge Stanley Sporkin refused to approve it. (Justice and Microsoft filed and won a joint appeal, which is why the whole thing is being relitigated now, though too late to help OS/2.) But no one who has studied the sad history of OS/2 would hold IBM free of fault in its demise.

In the late summer of 1997 I had occasion to talk with a number of IBM executives in a position of power over OS/2. Reiswig had been blown out of Personal Software Products in late 1995 and IBM entirely early in 1996, probably for having tried to get mere people excited about the product, though the reason given was the failure to port OS/2 to the microkernel that ended up being abandoned anyway. The news in 1997 was not good: They were advising OS/2 users--the ones who asked--to find themselves some other operating system. They didn't publicize this, and it wasn't covered, but OS/2 was dead. The IBM software execs were all excited instead, about something called "Workspace on Demand," a client-server arrangement not unlike the old mainframe-terminal days, only graphical. I never heard of this thing being actually adopted and used by anybody, though I'm sure they sold it to someone. But OS/2 was dead.

Its users, that Team OS/2 crowd, either continued to use what it had or migrated to Linux. It's surprising, the number of OS/2 alums now in the ranks of Linux users. Or maybe not--where else was there to go? You can pick them out, the veterans of the OS/2 war. They're the grizzled characters who have taken the place of the "I remember when we used punchcards" crowd in the newsgroups. (Well, okay, there are some major old Unix hands there, too, and their bona fides are at least as good.)

But now it's official. OS/2 has been pronounced dead. No rumor can revive it. A first, or at least the first since CP/M.

A sad day, for an operating system that deserved better. And that provided desktop innovations that no operating system has yet equaled.