Let's look at a few issues with the OP's analysis:
1) As others mentioned, there is a survivor bias.
2) Runway should be measured in months, not in millions. Size of funding to log size of exit is the wrong metric. Months of runway to IRR of exit is the better comparison.
3) I forgot what #3 was.
Even when the counter-argument isn't great, I still like the discussion.
[1] http://www.avc.com/a_vc/2013/09/maximizing-runway-can-minimi...