Has anyone read the “Focusing on the Long-term” paper from Google? [pdf]

Almost every company that's out there runs some form of A/B tests to measure the success/failure of their features. Typically, A/B tests span a few weeks (2-3). Is that timeframe enough to make critical decisions? What's your take on short lived A/B tests? I found the paper interesting. Has anyone tried this/something similar outside of Google?