Confidence type 2 is a new link between rationality and predictions about contrarian futures.
Start with a paradox: The concept of "overconfidence," in the tradition of Kahneman/ Tversky, implies that any attempt to try something that has never been successfully approached by someone else in the past is irrational. So breakthrough entrepreneurs would be irrational by definition.
Todd Zenger and I think that this is fundamentally wrong if entrepreneurs have reasons –a theory– of why things are different and they will succeed. Take an example: Space Travel is expensive. Say, SpaceX has a theory to make space travel cheap by making heat shields of SpaceShuttles cheap. They gather data and must conclude that the theory is wrong: heat shields remain expensive. Should they abandon their theory? We argue no! Confidence type 2 is belief strength in assumptions. The theory should not be abandoned, but rewritten: Keep confident in the "reusable spacecraft" assumption but drop the "cheap heat shield" assumption, if you can replace it with an alternative causal logic (like "turning around spacecraft and fire again to slow down at re-entry to the Earth's atmosphere").
Zenger and i develop the mathematics of how to link confidence type 1 (statistical confidence; Kahneman/ Tversky) and confidence type 2 to learn how to make so far unsuccessful things true after all.
Confidence type 2 also has implications for the foundations of AI (as confidence type 2 defines how to "frame" data and arrive at new priors), and model selection problems in Econometrics.