Five-Paper Series Formalizes Cohesion, Arrives at the IVP
Five Zenodo papers formalize software cohesion and derive the Independent Variation Principle, exposing gaps in common cohesion metrics.
A five-paper series now complete on Zenodo builds a formal chain from a rigorous definition of software cohesion to the Independent Variation Principle (IVP). It starts by defining cohesion as a 2k-tuple parameterized by partitioning rules, instantiates a concrete metric based on change-driver-assignment identity, proves four necessary and jointly exhaustive conditions for optimal modularization, links change-propagation minimization to total maintenance cost under an explicit coefficient condition, and synthesizes the chain into a single structural principle.
For engineers, the practical payoff is a critique with teeth: the series shows that widely used algorithmic cohesion metrics — method-call overlap, shared-field density — measure a structural proxy rather than cohesion as formally defined. It also justifies the common assumption that minimizing change propagation is the right modularization objective, by decomposing maintenance cost into access, alignment, cognitive, and domain-fixed components.
Notably, an independent preprint from last month reaches the same conclusion — that the change-driver-equality partition is the unique cost-minimizing modularization — via a completely different graph-theoretic counting argument on the element–change-driver incidence graph. Two unrelated derivations converging on the same partition strengthens confidence in the result. All five papers are freely available on Zenodo.