When I left Microsoft, it seemed like a “low-tech” high-tech company focused more on application innovation (aka integrated innovation) rather than technological innovation. Application innovation is associated with repurposing existing technology to new markets. It’s less risky than technological innovation, and it’s hard to argue with Microsoft’s financial success.
When I left, I saw Windows ME (16–bit technology masquerading as a 32–bit OS) on the horizon, uninspiring versions of Office, and a dated flat Windows API; the .NET Framework was nowhere in sight. I never installed Win9X on any machine I owned.
I recognized bad decisions early as they happened. One was the Office assistant; even Bill Gates called it “the f—king clown” the first time he saw it, but succumbed to Office’s BS rationalizations. (Employees internally and derisively referred to Clippy as TFC from the earlier epithet.) I actually agree with the principles of social interface, but it needed to be based on solid natural UI implementations (speech, natural language, etc); the cartoonish assistant in Bob, Office, and Windows discredited perfectly good research. Active desktop and shell-web integration were obvious kludges—hasty and bad decisions resulting from the Microsoft’s Internet panic of 1996. I didn’t think much of MFC or COM.
These days Microsoft produces less crap. Almost all of the early bad decisions have been reversed. Just recently, I noticed that Word 2007 killed off the Letter Wizard, which seemed like a misprioritization when it was initially added; instrumentation data probably revealed to Jensen Harris and gang that no one used the feature. As an example that bad ideas tend to hang around together, the Letter Wizard partnered with Clippy, so that whenever the user typed “Dear X,” a balloon appears, saying, “It looks like you are writing a letter,” and offering to invoke the wizard.
Windows Vista and Avalon have done a better job of integrating the best principles of the web—inductive UI, navigation, markup, etc— with the best features of Windows. The operative keyword is “principles.” Microsoft employees seemed smarter now with more advanced degrees; the higher salaries probably makes the company more enticing, despite the poor stock performance. The company snatches computer experts, who have made a name for themselves, like Erik Meijer, author of Haskell. There are now software architects and professional industrial designers in each product unit, looking at the product holistically and intelligently. Microsoft does seem to have topnotch technologists in charge of research. (No more dev managers of billion-dollar products who grimace at the term, object-oriented programming—nowadays, maybe FP.)
