Well at least I didn't see any evidence that says a good accountant would make a worse physicist than a bad accountant. So the assertion that people that are good at solving well-defined problems are not as good at (or at least has no correlation with) solving poorly-defined problems is not well supported.