Giant Missteps: How an IBM Attempt To Regain PC Lead Has Slid Into Trouble

Saga of OS/2 Software Shows Firm's Bureaucratic Ills, Hints of Problems Ahead

The Tension With Microsoft

By Paul B. Carroll
The Wall Street Journal

December 2, 1991

Long before International Business Machines Corp. launched its make-or-break decentralization plan last week, the giant computer company had a make-or-break plan to recapture its lead in the huge personal-computer market.

Now, that earlier effort in PCs is gathering momentum -- in the wrong direction.

"This has to be the greatest disaster in IBM's history," says George Colony, president of Forrester Research Inc., a consulting firm. "The reverberations will be felt throughout the decade."

How IBM managed to shunt a comeback train onto a sidetrack largely explains the growing urgency and widening sweep of Chairman John F. Akers's promises of change. Mr. Akers had already attempted a decentralization in 1988, rewritten a lot of rules through this summer's startling alliance with Apple Computer Inc. and cut tens of thousands of jobs.

However, his latest plan goes even further. It seeks to split IBM into numerous businesses and raises the prospect that those failing to measure up will be sold. It also seeks to prevent the sort of disaster that goes under the name of OS/2, IBM's operating-system software for personal computers. But the OS/2 saga shows how hard it will be for IBM to change.

Despite the technical overtones, OS/2 is really a management problem. It shows IBM's bureaucracy at its most arthritic and how Big Blue throws billions of dollars at problems that can't be solved with money. It shows how IBM clings to the idea that it must control every part of the market, even though the company must relinquish that heritage if it is to be as flexible as Mr. Akers's latest plan requires.

Nonetheless, IBM's developers can eventually solve even the toughest technical problems -- but sometimes too late to make a difference in the marketplace.

Asked about the prospects for OS/2, Scott McNealy, chief executive of workstation maker Sun Microsystems Inc., at first stares slack-jawed and then says: "OS/2? That one's already buried. There's grass growing over the grave."

That's not how IBM executives see it. They contend that the problems are temporary, that software projects take time and that OS/2 will take off when a new version becomes available next March. They say OS/2 is so good it may entice customers to buy the more-powerful computers needed to run it, lifting the PC business out of a rut. And they note that successful software is highly profitable.

"We are not in business to lose money," says Joseph Guglielmi, an IBM vice president responsible for OS/2 marketing.

The OS/2 problem takes the form of a race with Microsoft Corp. that IBM could have won but is losing by miles. Narrowly at stake is the $1 billion-a-year market for the operating-system software that controls the basic functions of personal computers. But whoever leads that market can shape the future of PCs and the "applications software" -- the programs that make them useful. That means a head start in reaching a projected $40 billion-a-year market with new machines and applications using the dominant operating system. Everyone else will play catch-up.

For IBM, the recent efforts to push OS/2 as hard as possible also represent an attempt to recover a multibillion-dollar investment, restore its credibility with bewildered customers and satisfy its corporate pride by winning a fight that is becoming extremely personal. Success with OS/2 might even convince investors that IBM can innovate enough with software that it won't soon have to make another painful announcement like that last week.

Even though IBM popularized the personal computer, sales of its OS/2 operating system are trailing Microsoft's rival Windows system by 15 to 1. Based on figures supplied by IBM and software-industry executives who have worked with it, IBM appears to have spent nearly $2.5 billion over the past six years developing OS/2 and applications that run with it. For all that, IBM has collected perhaps $100 million of revenue. Work on OS/2 alone has cost IBM some $850 million.

Because of the way IBM accounts for development costs, some $800 million of that $2.5 billion has yet to be treated as expenses and will weigh on earnings for several years. IBM is still investing some $350 million a year on the project.

Even those numbers, however, don't fully explain IBM's problems with OS/2. Linking OS/2 to the company's PS/2 line of personal computers also contributed to the decline in IBM's PC market share to about 20% today from nearly 40% in 1986, the year before OS/2 was announced. That cost: $8 billion a year in revenue.

In addition, OS/2 hurt IBM's credibility with software developers, who market researchers say spent more than $1 billion developing applications for OS/2. How much revenue did those applications bring in? "Zip," says Tim Farrell, president of Future Soft Engineering Inc., which made some OS/2 products.

The OS/2 problems go back to 1985, when IBM decided it had to fend off the clones. It chose to go its own way in PC hardware and software. It hoped to set new standards and put pressure on clone makers and to regain control of its destiny by doing the software on its own. But when IBM told Microsoft that their five-year relationship was over, Microsoft's cofounder, Bill Gates, appealed to IBM's chief executive, Mr. Akers. Mr. Gates won a reversal at a luncheon in June 1986.

The companies' ideas differed, however. Microsoft started out from a PC perspective, with Windows serving as the base. IBM brought mainframe ideas to the process -- including the idea that hordes of IBM software and hardware people should have a say in how OS/2 was developed.

Managing by committee has long hobbled IBM, where any member of the group can call a halt by saying he or she "nonconcurs." An allegorical memo circulating among IBM developers says IBM lost a rowing race to Microsoft because, some IBM task forces found, Microsoft had eight people rowing and one steering while IBM had eight steering and one rowing. The task forces' recommendation? The one rower should row harder. Eventually, he was fired for poor performance.

Although small, tightly integrated teams do best at developing software, IBM and Microsoft wound up with 1,700 programmers working at three sites on two continents. The result: a product far too complicated for mere mortals to use; far too big to run on many of the computers that IBM introduced together with OS/2 in April 1987; and far too revolutionary. Users wanted to move ahead in little steps and not have to trash all their current hardware and software.

To Microsoft, the problems with OS/2 persuaded it to return to Windows. Windows would provide many of the nice touches that make Apple's Macintosh so easy to use. It would also work together with -- and not replace -- the DOS operating system already on the tens of millions of IBM-compatible computers. Microsoft saw Windows as an interim step that would eventually lead customers to OS/2. IBM thought Windows would undercut OS/2.

In November 1989 in Las Vegas, Nev., the two companies scheduled a fateful dinner on the Sunday night before the big Comdex trade show began. They told the industry's 30 major software executives that they had patched up their differences: Windows would be positioned for less powerful computers, OS/2 for more powerful ones.

But Jim Cannavino, who runs IBM's PC business, apparently became uncomfortable as he heard people talk of how IBM had finally endorsed Windows. At the companies' joint press conference that Tuesday, he managed only anemic praise for Windows. Afterward, Microsoft executives bristled when they read that software rivals such as Jim Manzi, chief executive of Lotus Development Corp., interpreted the agreement as the emasculation of Windows. So Mr. Gates and Microsoft began putting more and more emphasis on pushing Windows hard.

When Microsoft introduced another version of Windows in May 1990, the software went from a steady seller to a juggernaut. Windows has now sold more than seven million copies, compared with only 800,000 for OS/2 -- and 300,000 of those OS/2 sales are in dispute. Microsoft says that IBM gave the copies away as part of a sale of memory boards and that they were never used. IBM disagrees heatedly. IBM also dislikes the way Microsoft abandoned OS/2 this year; it thinks Microsoft reneged on a commitment that would have ensured OS/2's eventual success.

"We were appalled," says Lee Reiswig, an IBM OS/2 development official.

As the companies began to go their separate ways, the personal frictions increased. IBM executives, accustomed to polite debate, resent what Microsoft's own executives call Mr. Gates's machine-gun style. When he disagrees, no matter with whom or in what setting, he fires questions -- What about this? What about that? -- until his opponent wilts.

People in the industry add that IBM feels it created Microsoft, which consisted of 32 people before IBM decided 10 years ago to use Microsoft's DOS operating system on the original IBM PC. And IBM executives say they are tired of reading about Mr. Gates's brilliance and billions.

Many industry executives say IBM should have cut its losses on OS/2 and moved on. In fact, IBM acknowledges that it did take a hard look at the project. But Mr. Cannavino says the company decided the original concepts were still sound. In addition, IBM had been pushing OS/2 so hard with its customers that it felt obliged to continue.

At the end of last year, IBM did disband much of a huge, related project, called OfficeVision, that had tried to be all things to all people across entire companies but that had hit technical and management problems. So, IBM's credibility in PC software was at an all-time low when IBM announced in April that it was going to make its biggest push yet on OS/2.

Its pride stung, IBM assembled customers, software developers and industry analysts for day-long sessions. People who attended say Mr. Akers and IBM President Jack Kuehler laid their reputations on the line, vowing that OS/2 would be delivered by year end. Even though many observers considered the goals too aggressive, these people say Mr. Akers told some senior IBM executives, such as Mr. Reiswig, that he would have their badges if they failed. And Mr. Cannavino growled to software developers that if they had any trouble with Microsoft, they should call him and he would personally take care of it. (IBM executives say some of the stories are exaggerated but aren't more specific.)

The soft-spoken Mr. Reiswig has taken to calling himself "the Blue Ninja" to show the ferocity with which IBM intends to push OS/2. At trade shows, he often wears -- over his button-down, white shirt -- a T-shirt with a Ninja warrior on it. He signs himself "the Blue Ninja" on IBM's internal electronic-mail network.

The pressure has won some converts, but IBM's fumbling attempts to stay on the cutting edge are damaging its credibility again. The IBM-Apple deal last summer made some customers wonder whether OS/2 wasn't just IBM's operating system "du jour," which they would be urged to replace in three years with something better. Earlier, IBM had spent some $60 million on a deal with Steve Jobs's Next Inc. for some avant garde software. Last year, IBM formed a software venture with Metaphor Computer Systems Inc. to address similar issues and spent $10 million on that. Neither plan went anywhere.

IBM further damaged its credibility this fall by admitting failure with its Desktop Software group, an attempt to buy the rights to leading-edge software packages that were to make IBM a factor in the PC software market.

Then came the big one. IBM delayed OS/2 from the end of this year until next March because it wanted to add more functions. Given its promise of a death march to get OS/2 out and its denials right up to the last minute that it was having problems, even some loyal IBM customers wonder what is going on.

"In the big scheme of things, the deadline they missed isn't a catastrophe," says Larry Bacon, a senior vice president at Travelers Corp. and long an OS/2 advocate. But "if they miss another one, well, how many strikes do they get?"

IBM has, in fact, made admirable progress on the technical side on OS/2. It has apparently even solved some problems so thorny that many considered them unsolvable. It also can count on many big customers to stay with it. If IBM meets its March deadline, it may also have nine months to a year in which OS/2 will have features unavailable from Microsoft, whose Windows NT will probably reach the market late next year or in early 1993.

A year might give IBM breathing room, and it is trying to exploit its advantage by lining up as many software applications for OS/2 as possible. But none of the applications on the horizon will give OS/2 the sort of advantage Apple's Macintosh had a few years ago when it had a monopoly on good desktop-publishing software.

Besides, Microsoft's success with software developers dwarfs IBM's. IBM talks of having more than 800 applications available next year; Windows already has 4,900. When Microsoft held a conference for Windows developers recently in Seattle, more than 2,000 showed up -- so many that, when Microsoft wanted them to walk half a dozen blocks to a special session, it had to get a parade permit from the police.

"DOS and Windows are going to own the desktop," said Fred Gibbons, chief executive of Software Publishing Corp., a former OS/2 supporter that has switched to developing for Windows.

Last year, a new trade magazine on PC operating systems called itself OS/2. Later, the name became Windows and OS/2. When CMP Publications bought the magazine this summer, it changed the name again -- to Windows.