"philipsu" offers an insider's perspective on Microsoft's corporate culture and why WinVista keeps getting delayed. That link is to a mirror since the original post was removed. My desire is not to knock Microsoft, whose software I generally like, but to point out that process is important, and that writing software is hard and the difficulty increases exponentially with the lines of code.
The real problem isn't Microsoft's culture! The modern programming paradigm inevitably leads to Gordian software. Solving this single problem, hidden in plain view from most computer scientists, could eventually be one of the greatest contributions from artificial intelligence to humanity.
Microsoft's primary problem is too much managerial overhead, with many managers bringing their own stuff to the table and pushing it on the programmers, often during the same meeting.
Well, the problems of large software systems is an interest of mine so I hope you don't mind if I mention that I can say with reasonable confidence that the Gordian Software guy doesn't know very much about computer science. The first clue is his idea that computer science is somehow bound by the assumptions it began with. John Backus, a computer scientist, was complaining of the same thing back around 1960 and it wasn't really true even then. There were Simula, Lisp, Snobol, APL and various other languages that looked nothing like the original languages. Since then we've had logic programming, constraint programming, pre- and post-conditions, fourth-generation languages, information hiding, multiple inheritance, subtyping, and a hundred other techniques for writing complex systems.
Of course it's possible that this guy has come up with a brilliant scheme that has escaped three generations of researchers, but I doubt it.