In 1988 I saved taxpayers million of dollars by talking my client, part of a government agency, out of its planned conversion to OS/2. Pundits were predicting OS/2’s success, but I saw no easy migration path from DOS, difficult integration into existing LANs, and no hurry. Besides, IBM’s decision to build OS/2 from scratch instead of creating “personal UNIX” struck me then, as now, as unnecessarily complicating its product line.

In 1990, a NeXT sales representative asked what it would take to bring some NeXT computers into our organization (I had rejoined the private sector by then). “It has to run a WordPerfect-compatible word processor, a Lotus-compatible spreadsheet, and log into our Novell servers,” I answered. It didn’t and I didn’t, seeing no easy migration path from DOS, difficult integration into existing LANs, and no hurry. Besides, I hated NeXT’s weird capitalization.

To invade an existing architecture, new technology must offer: (1) easy integration into existing environments; (2) a clear migration path; (3) a compelling advantage; and (4) attractive pricing.

And right now, as Bill Gates seems to dominate the known universe, invasion of the desktop OS is possible for the first time in years. Four disparate factors lead me to this conclusion.

Factor #1: Dissatisfaction with current PC operating systems. The market leaders are all awful. Windows/95 is a kludge, Windows/NT is flabby, OS/2 only runs on Intel and has no applications, and the Mac OS lacks memory protection and preemptive multitasking. And they’re all hard and expensive to manage.

Factor #2: Openness to other solutions. The Network Computer, while a bad idea, has created a crack in Microsoft’s mindshare, and mindshare – the perception that a company is now and will be an important player – is all that matters. If you have mindshare, marketshare will follow; if you don’t, marketshare is irrelevant. If you don’t believe me, just look at Novell’s slide into irrelevance.

Factor #3: Bill Gates’ imitation of Napoleon. The similarities are compelling: while physically unprepossessing, Napoleon was and Gates is a brilliant strategist and tactician. Napoleon fought a two-front war and lost his empire. Gates has instituted repressive laws “at home” (Microsoft’s ridiculous new desktop licensing terms, reported in several recent issues of InfoWorld) and has engaged its enemies (everyone else) on the network OS, Internet, and DBMS fronts. That’s a passle o’ wars to fight all at once.

Factor #4: Steve Jobs and NeXTStep at Apple. While Gil Amelio reportedly is a shrewd businessman, he’s not the visionary Apple needs to reclaim industry mindshare. Steve Jobs can grab industry mindshare while the Apple name can give NeXTStep instant industry credibility.

Put these factors together, and imagine Apple does everything right for a change. Imagine it ports NeXTStep to the PowerPC; makes the transition from the current Mac OS to MacNeXTStep as easy as it made the transition to the PowerPC a few years ago, building backward compatibility into MacNeXTStep so current Mac applications will run well; keeps the Mac’s legendary ease of set-up intact; makes it easy to port applications written for other UNIXes to MacNeXTStep; builds in manageability (easy to do with UNIX); and matches current Wintel price points. Oh, and demonstrates betas by mid-year and ships by 1998.

You’d have a new desktop architecture with everything going for it: price, integration, migration path, compelling advantages, and industry credibility.

How does this affect you today? You’re writing client/server applications right now, and you have to make decisions on the desktop platforms you’re writing them for. Every new application makes transition to a new architecture more expensive, no matter how attractive the technology may be on its own merits. So what should you do?

Watch closely, make short-term, pragmatic decisions, and invest in flexible, portable application architectures. The world is about to become even more interesting.