The old style: Some 'expert' says: 'Run 100 trials.' His opinion: the new system is same as the old one. Results: 50% of the time it's a wrong conclusion, and very, very wasteful of test resources. Far too many tests, or far too few !
New style: Integrate correct Statistical Methods into T&E sign-off. Forget buzzwords (like 'DOE'). Construct appropriate --'statistically defensible'--hypothesis tests. Be right at least 95% of the time. Understand, mathematically, the tradeoff between size of test and risk of being wrong. (Used effectively by successful industries-why not us ???)