2. More redundancy can be revealed by expanding scope.
3. The ideal scope size is what can be reasoned about independently and specified independently from the rest (there's a global optimisation problem here, of trade-offs between the scopes of adjacent parts).
4. The scope for measuring redundancy can also be expanded, to include all things reasoned about and specified by a person. This enables us to account for specification techniques which are complex in themselves (in an absolute sense), but which improve simplicity overall, by being applicable in many situations (generalised) - they can remove redundancy across unrelated domains by factoring it out.
an information-theoretic perspective