I was going to retort to your statement about changing failure rates saying that's not side of the equation we're talking about. If we take this as the number of startups we create as:

   numberOfSuccessfulStartups = startupAttempts * successRate

Then Jon Stewarts observation is by lowering risk failure we raise the startupAttempts. Not only do we raise the attempts, but we can recycle the failures to drive startupAttempts up more resulting in a higher number of successful startups is his hypothesis. Adding health benefits, tax credits for loses, etc drive up startupAttempts by lowering risk.

However, your observation that making things cushy for entrepreneurs could lower successRate because you don't have to hustle as much. Sort of a if I fire this gun above your head while you work maybe you'll work harder argument. I say that in some jest because I think there is some merit to your observation just how much I don't think either of us know.

The one counter I can provide to your counter is that lowering risk might have the same effect as lowering incentives for results. Basically support from this work on how incentives can lower success if they are too high. What I'm saying is maybe if risk is too high fewer people attempt and even fewer people succeed because there aren't enough original ideas to copy for scale.

http://carmine.se.edu/cvonbergen/Incentives.pdf

(Couldn't find a better link.)

I think both hypothesis have their points and we need more data to support them.