As I struggle through conceptual models for the next generation of our software, I'm realizing that what we traditionally think of as "applications" are mature. However, the file system, and most of the ways we've developed to look at it, are lousy ways to represent data, and most new applications will be about developing better ways to represent what we think of now as "the desktop". If the open source and Linux folks want to leapfrog Microsoft, they'll stop trying to duplicate .NET and start thinking about how to integrate rows in tables and files in the filesystem and how that information flows around, and building higher level frameworks to move that data, rather than (as Microsoft is currently doing) trying to reinvent the wheel for which we already have reasonable protocols. At least that's what I'm looking at.