Do entrepreneurs who "fail early" have a better chance of future success?

James Surowiecki has a terrific column on entrepreneurship in the New Yorker this month.   In the article, he asks why so many entrepreneurs continue to launch start-ups despite the low probability of success.   Here's one excerpt that I found interesting.  In this excerpt, Surowiecki questions whether an early failure leads the type of learning that can generate future success: 

There’s a widespread tendency to treat failure as a badge of honor: “Fail fast, fail often” is a familiar mantra in Silicon Valley. There’s now a regular FailCon, where people come to hear other entrepreneurs tell about the hard times they endured and about how starting a business and failing actually makes you more likely to succeed in the future. It’s a comforting message, but the evidence suggests that past failure really just predicts future failure. A 2009 study of venture-backed firms found that entrepreneurs who had failed in the past were not much more likely to succeed in new ventures than first-time entrepreneurs were—some eighty per cent of those who had failed before failed again. A later study of more than eight thousand German ventures came to an even grimmer conclusion: founders who had previously failed were more likely to fail than novices.