Where Microsoft fails

Since the early days of Novell's PC networking, it was clear that Microsoft didn't have a clue what to do about networking. While Novell's networking simply worked seamlessly with MS DOS, Microsoft's Lan Manager was a clumsy black art. It only became popular when it was integrated into Daytona (which became Windows NT, which became Windows 2000, which became Windows XP, ... 2003, Longhorn ...).

Microsoft's next attempt was in distributed objects. The basic technology was Distributed Common Object Model, DCOM. This was a big fat ugly technology, but elegant when compared to other Microsoft services such as OCX, OLE2. They've continued to believe that network speeds will improve with CPU speeds. To everyone else, this is obviously never going to happen.

The next attempt at networking was to provide a Multics style computing service, where all the important apps are distributed on Microsoft servers. You'll then have to pay a Microsoft utility bill to access your own stuff. The technology to make this happen? .NET. After some years of dabbling with .NET, Microsoft still don't offer a platform and tools to develop the latest .NET 2 apps and the existing .NET 1.1 apps.

Enter Google. Have you seen Google Mail, Calendar, Spreadsheet? They're bloody good applications. They're easy to use, and slick to look at. And of course, there's searching on nearly everything. If Google brings out an online word processor, it'll be all over. The Operating System will just be a tool to access the world wide internet and run viruses.