All I'm claiming is you pour the innovation out of Google Corp Hq into a bucket and, um, how about MIT Media Lab into a bucket, and measure the two buckets on a scale. I feel MIT wins handily. But for the sake of argument lets say Google kicks the MIT Media Lab's butt to the tune of 10x as much innovation. There is still a slight problem in that the MIT lab is about 100 people (to one sig fig) and Google Corp is just under 60K according to a Google search. Assuming similar quality of "human resources" Google should be consistently producing 600 times as much innovation per year as MIT Media Lab. Maybe when you factor in Google's company purchases, after which all innovation at the purchased company traditionally ceases...
Isn't Google fundamentally on the scales of justice more of a profitable advertising sales boiler room than a source of innovation?
If as a company, its mostly about being a sales boiler room, then it should look like a sales boiler room, shouldn't it? Perhaps there is less inconsistency between what is observed vs theory after all.
The essence of making workspaces work is minimizing distractions. Any layout can work if you take the chatter ELSEWHERE.
It takes a while to get a space ready, and it's also not easy to just expand around existing offices when they're already quite large.
Admittedly, much of this would probably be solved if they were more OK with remote working, as there's a high emphasis on in person collaboration. I suspect they've done some studies on this matter though.
I work for Google but opinions are my own.
I know I got used to a lower productivity after spending some time in open offices. Then one day I was sitting in a private office again and felt my productivity surge and felt: "WOW! This is the way programming should feel like!"