[nradov]

The classic business management book The Innovator's Dilemma specifically recommends moving groups working on potentially disruptive new products into separate facilities. That helps them avoid distractions and makes it harder for "corporate antibodies" to kill the new initiative. Most people think that the IBM PC project succeeded in creating a microcomputer that disrupted the mainframe and minicomputer markets at least partly because they had a separate office in Boca Raton instead of at HQ in Armonk.

http://claytonchristensen.com/books/the-innovators-dilemma/

Of course having a separate facility doesn't guarantee success, and can certainly cause other problems if poorly managed.

[WalterBright]

The classic of this is the Lockheed Skunkworks. It's success has never been repeated - lots of companies try to create a skunkworks clone, but try to fix it, and wind up thereby breaking it.

https://en.wikipedia.org/wiki/Skunk_Works

[marcosdumay]

Google+ would never be disruptive to Google. It is in a completely different market from anything else, it wouldn't have to destroy any other Google project to succeed.