Regarding the final bullet point of doing a second validation, the sample size should be bigger right? Because of the tendency for winners to coincide with +ve

Regarding the final bullet point of doing a second validation, the sample size should be bigger right? Because of the tendency for winners to coincide with +ve random effects, you will choose a larger experiment size and expect to see a lesser result.