Cohesion and Knowledge Embodiment
This page covers a central result of the cohesion theory: the knowledge-embodiment theorem. It is placed here, under design, rather than under the IVP core theory, because it is a theorem about cohesion, not a theorem of IVP itself.
1 Cohesion, briefly
Cohesion is correctness relative to a partitioning rule. Under any modularization principle that groups elements, a module has two properties:
- Purity. The module contains no element from outside its correct class.
- Completeness. The module contains every element of its correct class.
A module that is both pure and complete is maximally cohesive: it holds exactly the right elements and nothing else. When the partitioning rule is the change-driver rule, this recovers the IVP notion of maximal cohesion — but cohesion itself is the more general idea, defined for any rule, not only for IVP’s change-driver rule.
2 The knowledge-embodiment theorem
The theorem connects this structural measure to an epistemic one.
A module has maximal cohesion under a rule if and only if it embodies exactly the knowledge its -class must hold — every knowledge piece its class requires, and no knowledge piece its class must not hold.
In words: maximal cohesion is maximal knowledge embodiment. Get the structure exactly right, and the module’s knowledge is exactly right, with no separate effort.
This is why the result lives in the cohesion theory. It is a statement about cohesion — it says what maximal cohesion means epistemically. IVP supplies one particular partitioning rule (the change-driver rule); the theorem itself is about cohesion in general.
3 What maximal cohesion guarantees
A maximally cohesive module has four structural properties.
- The module is epistemically self-contained. No knowledge its class must hold resides outside; no foreign knowledge resides inside.
- The change radius is bounded. When a change driver governing the class’s knowledge activates, every affected element is in one module.
- The module is independently verifiable. Correctness is checked against the class’s own knowledge, with no external consultation.
- The module is independently replaceable. A replacement need only embody the same knowledge and answer to the same change drivers.
These are the structural conditions that underlie the benefits of modular programming under information hiding.
4 What it does not guarantee
The phrase “maximal cohesion is correct embodiment” is easily over-read as “maximal cohesion is total correctness.” It is not.
A system has three distinct properties that this phrase blurs:
- Placement — the right knowledge sits in the right modules. This is what cohesion measures.
- Faithful embodiment — each piece present is realized correctly. This is functional correctness.
- Dependency parsimony — the dependencies among elements are the ones the purpose requires.
These three are independent. A module can score maximal cohesion and be wrong (one element computes its piece incorrectly). A module can be maximally cohesive and faithful yet carry an unneeded dependency. A system can be entirely faithful and badly placed.
Cohesion governs the cost of achieving and maintaining correctness — the scope that must be read and the radius that must change — not whether the system is correct. A correct system may be incohesive and expensive to keep correct; a cohesive system may be incorrect and cheap to repair.
5 The knowledge floor
At maximal cohesion across all partition-inducing rules, every knowledge piece resides in exactly one module, and the system’s relevant causal domain knowledge is partitioned across exactly as many modules as there are classes. This is a theoretical lower bound on module count — a structural minimum, not a practical recommendation. Subdividing within a class for readability or team structure is freely available.
6 Where this sits in the theory
The knowledge-embodiment theorem is proved in the published paper on the nature of cohesion [1], where cohesion is defined as correctness relative to a partitioning rule and the equivalence between maximal cohesion and maximal knowledge embodiment is established.
IVP and cohesion are related but distinct. IVP is the principle that groups elements by change-driver assignment. Cohesion is the more general measure of correctness relative to any partitioning rule. The change-driver rule is one instance of that general idea — the instance IVP is built on.
The formal development. The definitions, theorems, and proofs behind this page are published in:
- On the Nature of Cohesion: Cohesion as Correctness Relative to a Partitioning Rule — Zenodo, 2026 — doi:10.5281/zenodo.21703193