And if the source of that framework is not (part of your definition of) science, how do we know whether that framework is true?
That was sort of my point: Poppers criterion is nice, but only works for a small subset of (natural) science - and even there has to face criticism because it is simply too naive for many edge cases.