> Except that the new iteration is pretty much making random changes.

E.g., avoiding ivory tower planning and instead using empirical, in-the-wild A/B testing.