> I hope that at some point most people will acknowledge implicitly that the fewer the number of measures the more probable is that it is a simplification that hides stuff.

But the larger the number of measures, the more free variables you have. Which makes it easier to overfit, either accidentally or maliciously.