Refusing to concede defeat in the face of daunting opposition, the International Business Machines Corp. is once again trying to convince the world that OS/2 is the best operating system for the current generation of personal computers.
This time it may be right, though right doesn't always make might in the software business. At a technical symposium last week in Austin, Texas, the home of its new personal software products division, IBM introduced what is by far the most impressive version yet of OS/2.
Unlike other personal computer operating systems, OS/2 version 2.1 allows users to operate DOS, Windows and OS/2 software at the same time. That means that users don't necessarily have to throw out their old applications to take advantage of new features.
Unlike Windows, which runs DOS software slowly, OS/2 can run DOS and Windows applications just as well, and perhaps faster and more reliably, than their native operating systems.
Version 2.1 of OS/2 (suggested list price, $249 for the next 90 days; current street price, less than $100) is the first personal computer operating system to tap into the potential power of 32-bit microprocessors, including the Intel Corp.'s 386, i486 and Pentium chips. The 32-bit microprocessors have been around for years, which shows how far software lags behind hardware.
Advantages of 32-bit
The advantages of 32-bit software are often described by such esoteric terms as pre-emptive multitasking, multiple threads and use of the full 32-bit instruction set and extended registers of the Intel processors. What that all boils down to is being able to do more, faster, with less effort.
The user can work on a spreadsheet file, while the computer gathers and sends electronic mail in the background, for example, or work on one document while the computer is repaginating another.
IBM also outlined what appears to be a clear and solid strategy for guiding its corporate customers into the world of distributed computing, where the centralized systems of mainframes and dumb terminals are replaced by interconnected webs of powerful, small computers.
IBM has designed OS/2 to span the range from yesterday's 16-bit applications to tomorrow's object-oriented programs, which use building blocks of computer code to expedite customized applications, and to connect everything from big mainframes to small hand-held communicators.
But even with a superior product, a clear strategy, a big head start over one perceived 32-bit rival -- Microsoft Corp.'s forthcoming Windows NT operating system -- and a record of success against the sophisticated Unix operating system, IBM still faces a long uphill battle.
OS/2's troubled history
OS/2 was once heralded by Microsoft as the successor to DOS -- before Microsoft decided to ride Windows into the future and dumped OS/2 into the hands of its estranged partner, IBM.
It took IBM many years to learn how to develop PC software on its own. Even when it finally came up with a decent version, OS/2 2.0, it bumbled the marketing.
"You want to know how to screw up a product launch?" one IBM official said, in remarks so candid they almost amount to breast-beating. "Announce the product 20 minutes after you finish development, price it aggressively but then don't send it to the stores right away, sell it directly to customers and tick off your dealers and have no major business applications for it."
IBM learned from that marketing disaster, but Microsoft's marketing juggernaut leaves no margin for error. Windows 3.1 captured the desktops and created a halo for Windows NT, and an estimated 24 million copies of Windows are in circulation, compared with 2 million copies of OS/2.
Although IBM insists that it is "in to win," many analysts believe OS/2 cannot overcome Microsoft's momentum on the desktop, even though the newest version of OS/2 is technically superior to Windows 3.1. Many business customers are planning to move Windows NT, preferring the promise of NT to the track record of OS/2.
"I do a lot of research as far as what direction the market is going, on connectivity and the type of support that the market is going to bring to me as a developer," said Jim Ray, a programmer analyst for the McDonnell Douglas Corp. "OS/2 just isn't in the cards. It's not that I have anything against IBM, it's just that I see a lot of support going to NT."
Windows NT, which is aimed at the powerful network PCs called servers, demands more memory and processing power than are found on office desktop computers today. The program has been delayed since Microsoft first announced it. And corporate customers are unlikely to entrust their critical business applications to NT until it has been tested thoroughly, which means it is not expected to make significant inroads into the corporate market until next year.
"We've outscored NT, 2 million to zero, and we've outsold all the different types of Unix over the last nine months," said Wally Casey, director of marketing for IBM's personal software products division.
The marketing war
Yet even though NT has not yet been deployed in businesses, Microsoft has been winning the marketing war.
How frustrating is this to IBM? So much so that in recent ads touting OS/2, it has referred to NT as "Nice Try."
Donald A. DePalma, an analyst for software strategy at Forrester Research Inc., said: "IBM is getting a lot more aggressive. If they can leverage the new aggressiveness with the obvious technical strength, they may be able to get into some sites where NT is starting to smell a lot like OS/2 version 1.0."
What both NT and that first version of OS/2 have in common, DePalma said, is coming late to market and missing some features that were originally promised. And, he noted, Microsoft does not have IBM's experience in supporting heavy-duty applications and vast corporate information systems.
Although no one is predicting victory for OS/2, some observers say it has already found some measure of success among corporate software developers, whose interests are not always parallel to those of commercial software developers.
A commercial software developer wants to pick the one horse that will win the race; the corporate software developer wants to find the best harness for all the different horses in the stable, one that will enable them to work together most efficiently.
"You have to define what success is," said Esther Dyson, publisher of Release 1.0, a newsletter that analyses trends in the software industry.
"OS/2 will gain some share of the market. But will it win? No. You don't win in this game; you can only stick around and play another round. It's not going to wipe out Windows NT and Unix."
At some point while OS/2, Windows NT and Unix fight for position in the current round, "people will wake up and realize that the real action is with Taligent," Ms. Dyson said, referring to the operating system being developed jointly by IBM and Apple Computer Inc., with expected delivery in 1995. At that point, the survivors will get to play in the next round.
"Sure, they won't last forever, but nothing does," Ms. Dyson said. "They will die before one wins, although probably Windows NT will win in terms of having a lot of market share. Unix also has a big place, especially in business applications and running database servers and stuff like that."
(Peter Lewis works out of the New York Times' Austin, Texas, bureau:  328-8258.)