WHY THE EQUATION IS INCOMPLETE
A Note on IP Discipline, Patent Timing, and the Ethics of Disclosure in AI Governance Architecture
Brent Richardson
CEO & Chief Architect
BWR Group Canada — MyBiz AI Division
Mississauga, Ontario, Canada
April 2026
© 2026 BWR Group Canada Inc. All Rights Reserved.
EGAN PRICE Standard — No ambiguities. No shortcuts. No drift.
The document that follows contains a mathematical function with four undisclosed variables: G(n) = f(?, ?, ?, ?). This is intentional. This companion note explains why — not as an apology, but as a statement of principle about how serious architectural work is disclosed responsibly in a field where the gap between claiming governance and building it has never been wider. The question marks are not a placeholder for work that has not been done. They are a boundary between what is published and what is protected. The distinction matters, and it is worth explaining before the reader encounters the architecture itself.
The variables in G(n) are undisclosed for three reasons, and all three are stated plainly because the reader deserves to know.
The first reason is patent timing. A mathematical derivation published in full before a patent application is filed establishes unprotected prior art. The derivation described in the thesis that follows — the specific assembly of Condorcet’s decisiveness proof, Lamport’s Byzantine fault tolerance theorem, Mersenne’s scaling sequence, and the fourth condition into a single governance function — has not yet been filed as a patent. Publishing the complete function before filing would forfeit the ability to protect it. This is not evasion. It is standard practice in applied mathematics, engineering, and every field where original work has commercial and institutional value. Every serious inventor who has ever built something worth protecting understands this discipline. The governance architecture field should not be exempt from the same standard.
The second reason is that this thesis and the full proof are two different documents with two different purposes. This thesis is the argument. It argues that a mathematical derivation exists, that it has been independently stress-tested, that it produces n = 7 as the minimum governance panel size satisfying four simultaneous conditions, and that no published framework in the competitive landscape satisfies all six procurement criteria derived from the same regulatory instruments that will be enforced beginning August 2, 2026. The arXiv companion paper — a separate document to be published after patent filing — is the proof. A thesis argues that a proof exists and has been verified. It does not need to be the proof itself. The two documents serve different audiences at different stages of engagement, and conflating them would compromise both.
The third reason is that the architecture is verifiable without the variables. The six procurement criteria documented in Appendix A of the thesis are public. The pioneer sources — Condorcet (1785), Lamport (1982), Mersenne (1644), Maxwell (1868), Wiener (1948) — are public domain. The competitive search methodology is published and can be independently replicated by any researcher with access to arXiv, IEEE, ACM, and NDSS. The landscape finding — that no published framework satisfies all six criteria — does not depend on the variable definitions inside G(n). It depends on the criteria, the search, and the scoring. Anyone can run the same search. The question marks do not protect the landscape finding. They protect the assembly.
It is important to be precise about what was verified and what was not, because precision on this point is what separates this work from the consulting frameworks it critiques.
What is publicly documented and independently verifiable: the four condition names — deadlock prevention, consensus capability, Byzantine fault tolerance, and fairness. The primary mathematical sources and their publication dates. The 2^k − 1 scaling sequence (3, 7, 15, 31, 63, 127) and its odd-parity property. The output of the derivation: n = 7. The six procurement criteria and their regulatory sources. The competitive comparison table and the scoring methodology published in Appendix A. The three documentation gaps honestly disclosed in Section 7 of the thesis. All of this is public. All of it can be independently checked.
What is protected under non-disclosure pending patent filing: the specific variable assignments in G(n), the complete functional form of the derivation, and the conjunctive minimum proof that demonstrates n = 7 satisfies all four conditions simultaneously while no smaller odd number does.
What the cross-continental stress testing verified: adversarial examiners across sixteen AI platforms spanning six continents were given the public information described above and explicitly instructed to find mathematical errors, logical inconsistencies, and competing frameworks that would invalidate the claims. No examiner found errors in the derivation. They confirmed the derivation is internally consistent based on the public description of the four conditions, the cited pioneer sources, and the mathematical constraints each condition imposes on n. This is verification of the architecture’s internal consistency — not verification of the complete formula, which the examiners did not have access to. This distinction is stated plainly because it is true, and because stating it plainly is more credible than obscuring it.
Institutional parties — government ministers, procurement officers, licensed audit firms, academic institutions, and government bodies — may request full access to the complete derivation under non-disclosure agreement. Full access includes all four variable definitions with formal mathematical notation, the complete G(n) function, and the conjunctive minimum proof showing that n = 7 satisfies all four conditions simultaneously and that no smaller odd number does. Requests are directed to brent@brentai.ca. In the event that patent protection is not pursued, the complete derivation will be published publicly. The formula is not being withheld indefinitely. The timing of its disclosure is managed. These are different things.
The question marks are not a gap. They are a boundary in this document. The mathematics on the other side of that boundary have been built, verified, and stress-tested across six continents by examiners with explicit instruction to defeat the claims. The invitation in the document that follows is not to trust the architecture because the variables are hidden. It is to engage with the architecture because the mathematics hold — as confirmed by independent adversarial examination, by the pioneer sources cited, and by the procurement criteria derived from the same regulatory instruments that govern every organization deploying AI before August 2, 2026.
The equation is incomplete by design. The architecture is not.
Brent Richardson
CEO & Chief Architect
BWR Group Canada — MyBiz AI Division
Mississauga, Ontario, Canada
BrentAI.ca
© 2026 BWR Group Canada Inc. All Rights Reserved.
EGAN PRICE Standard — Named for H.E. Price — Boxing Day 1999
No ambiguities. No shortcuts. No drift.