Interesting conversation that could possibly takes hours to get to the bottom of. The MSFT bet was clearly a poor one at best.
I disagree.
From what I saw while I was there, Google scales through good process and discipline. I'm sure there are hacks that I didn't see that make things more efficient. However if you are truly dependent on scary hacks just to keep running, then you will fall apart when your scale increases, those hacks fail, and nobody knows how to make it work. Which is why Google is so big on scaling through good process and discipline.
As for why Microsoft is possibly a poor choice, there is a lot more institutional experience out there for making Linux scale than making the Microsoft stack scale. Furthermore Microsoft forces you to pay a hefty licensing tax for very unclear returns. So you're paying money to sail into uncharted territory and hope you can make it work.
But all that said, if they have architectural problems, odds are very high that the biggest problems are in the software that they wrote rather than the technology stack they use.
While Los Angeles doesn't stereotypically have good developers, good developers exist here. Demand Media and Google have 300+ people offices here. I suspect the problem is, either MySpace didn't hire good developers/architects, or MySpace hired them but management didn't listen to them.
Does starting a successful company depend on having a geographical talent base, heavily grouped together?
What does that mean, specifically? I've found it is a great stack, and the biggest problem it has is public perception :)