Regardless of the applicability to pg, I think this is a smart conversation. When evaluating hypotheses, we have a cognitive bias in favor of past success, when past failure might be the more informative signal. Planet Money did an interesting podcast with the author of this book:

http://www.amazon.com/Adapt-Success-Always-Starts-Failure/dp...

discussing just such a hypothesis:

http://www.npr.org/blogs/money/2012/09/04/160555540/episode-...