I had an interesting thought today as I was surfing the net about business culture and eXtreem Programming. I ran across a news item about Guido van Rossum finally finishing up a his first project since getting hired at Google.
Google has done very well as a web business because, not only do they have really smart people working for them, but, as a web software business, they have an unusual focus on human networking. That means they have a large number of technically capable alpha testers. Internally they can test their own software, and get good feedback from other software developers. Every project they test can be tested internally by anyone interested, and anyone interested will give useful feedback (as a consequence of being a software developer themself). It also means that Google understands what it means to work together on large software projects. It is principally this fact that has both launched Google ahead of Yahoo! and M$ and kept them there. (Internally they exploit many prediction market ideas from hiring process to who-knows-what-else. Who’s to know what other crazy brilliant ideas they use for self-regulation.)
As a further consequence of centralizing all that intelligence onto a single campus, they also generate good ideas. Ideas which can easily be bounced off other smart folks. Ideas which, if interesting, can be tested internally on a bunch of smart folk. As long as Google can avoid developing an internal bureaucracy (i.e. avoid over-emphasis on formality) which would eventually stifles new ideas, they will continue to be a great company.