Derivable Architectures — Volume II

Paper 018 — Canonical Derivations

Research Program: Derivable Architectures

Volume: II — Structural Theory

Document: papers/018-canonical-derivations.md

Status: Draft 0.1

Authors

Eduardo N. Hering
OpenAI ChatGPT


Abstract

Previous papers established that multiple derivations may belong to the same equivalence class while differing in implementation history.

This naturally raises a fundamental question.

Can one derivation be regarded as structurally preferred?

This paper introduces canonical derivations.

Rather than representing a mandatory implementation procedure, canonical derivations provide reference representatives for equivalence classes, enabling comparison, optimization, verification, and communication without altering architectural identity.


1. Observation

Equivalent derivations need not possess equal practical value.

Some are

  • easier to understand;
  • easier to verify;
  • easier to automate;
  • easier to maintain.

Yet all remain architecturally correct.


2. Observation

Engineering practice frequently converges toward preferred implementation procedures.

These preferences improve engineering efficiency.

They do not redefine the architecture.


3. Definition

Definition 1 — Canonical Derivation

A canonical derivation is a distinguished representative of an equivalence class of derivations.

Its distinguished status arises from explicit structural criteria rather than from architectural necessity.

Canonicality therefore concerns representation, not correctness.


4. Proposition

Every canonical derivation is correct.

Not every correct derivation is canonical.


Proof Sketch

A canonical derivation belongs to an equivalence class.

Membership requires architectural correctness.

Canonicality merely selects one representative from that class.

Therefore,

correctness is necessary, but not sufficient, for canonicality.


5. Observation

Different engineering objectives may produce different canonical derivations.

Examples include

  • minimum implementation effort;
  • maximum modularity;
  • maximum traceability;
  • maximum automation;
  • pedagogical simplicity.

Canonicality therefore depends upon explicitly stated optimization criteria.


6. Definition

Definition 2 — Canonical Criterion

A canonical criterion is the rule used to distinguish one derivation from the remaining members of an equivalence class.

The criterion is external to architectural identity.

It must therefore be declared explicitly.


7. Observation

Architectural correctness is objective.

Canonicality is objective only after its criterion has been fixed.

The distinction prevents confusion between engineering preference and architectural necessity.


8. Proposition

Distinct canonical criteria may select different canonical derivations within the same equivalence class.


Proof Sketch

Suppose one criterion minimizes implementation effort while another maximizes modularity.

The corresponding optimal derivations need not coincide.

Both remain architecturally equivalent.

Therefore,

canonicality is criterion-dependent.


9. Observation

The ICA implementation illustrates this distinction.

One implementation sequence may be preferable for manual drafting.

Another may better support automated document generation.

A third may maximize legal review efficiency.

Each realizes the same institutional architecture.


10. Definition

Definition 3 — Canonical Family

A canonical family is the collection of canonical derivations selected under different explicit optimization criteria.

Canonical families organize engineering strategies without modifying architectural identity.


11. Corollary

Canonical derivations separate optimization from architecture.

Optimization changes representation.

Architecture remains invariant.


12. Observation

Canonical derivations play the same role that normal forms play in many scientific disciplines.

They simplify comparison by providing standardized representatives while preserving underlying structure.

No assumption is made regarding uniqueness.


13. Conjecture

Certain classes of derivable architectures admit naturally preferred canonical criteria that emerge from the architecture itself rather than from engineering practice.

Determining when this occurs remains an open question.


14. Open Questions

  • Under what conditions is a canonical derivation unique?
  • Can canonical derivations be computed algorithmically?
  • Can canonicality be preserved under composition?
  • How do constraints influence canonical representatives?
  • Can multiple optimization criteria be reconciled?

Conclusion

Canonical derivations provide standardized representatives of architecturally equivalent implementations.

They distinguish engineering optimization from architectural correctness while preserving the structural integrity of the derivation space.

This separation prepares the way for studying the complexity and minimality of derivable architectures.