Getting Out the Crystal Ball

Computerworld - Every year after the dust of the holidays settles, I like to wipe off the old crystal ball and take a peek into the future of IT. Before I get started with this set of predictions, allow me to boast a moment about my record. My predictions may not pan out as quickly as I expect, but they usually come to pass.
I began talking about Linux in 1995. In 1996, I was one of the few to predict that it would become a mainstream server operating system. In 1997, I predicted that it would supplant Windows NT as the future server platform of choice when others were still saying it had a snowball's chance in hell. When Oracle gave a thumbs up to Linux in mid-1998, IBM still insisted that it had no intention of supporting the platform. Nearly five years later, IBM adores Linux, now the fastest-growing server platform. Score a big one for the Petreleymeister.
Network Computing Inevitable
This year's first prediction isn't a new one, but an extension of a previous prognostication that hasn't yet been fulfilled. As unlikely as I'm sure it seems, I predict that the era of network computing is still inevitable.
Granted, it seems very unlikely. I'd love to say that the death of the network computer has been greatly exaggerated, but it hasn't. It's difficult to find anyone discussing the concept of a network computer, let alone locate someone actually using one. Strictly speaking, every X Terminal qualifies as a network computer, but you won't find many of the cheap, Java-based desktops that Oracle, IBM and Sun had hoped would flourish.
Nevertheless, I predict that the Java-based network computer will rise again - and perhaps fall again. But it will eventually be a smashing success.
Java on the Client
My next prediction is a prerequisite of the last one and one of the reasons I still feel bullish on the network computer: The next two years will see a huge revival of interest in Java on the client. This will catch many people by surprise, especially after client-side Java was soundly trashed by embarrassing failures like Java-based WordPerfect Office.
One indication that Java will enjoy success on the client is that the platform neutrality of Java has improved dramatically over the past few years.
Java's performance has improved, too, but not so much that feature-laden Java programs run as fast as equivalent applications written in C. That performance penalty isn't bad news, though, at least for companies desperate to find a good reason to convince you of the need to