I've been thinking the same as I read this discussion.

My current best theory is that the software industry is overflowing with money, and it's still more beneficial to try to optimize for output quality and quantity (i.e. hire better software engineers no matter the cost, bring people into the same locality instead of hampering progress by splitting work into different timezones), instead of trying to cut costs to improve profits.

Once the growth finally ends, outsourcing to cheaper places might become a very attractive option.