EXHIBIT 1 · CRITICAL
Corrected: March 23, 2026
The 1600-Watt Brain
Claimed: The EARNED LIGHT framework included a calculation that the human brain at peak conscious operation could consume approximately 1,600 watts.
True: The human brain consumes approximately 20 watts total. 1,600 watts would cook it. The figure was derived from a mathematical extrapolation that failed basic biological sanity-checking.
Changed: All absolute wattage figures removed. Replaced with a relative-unit model: C(Ψ) = (Ψ / Ψ_baseline)² with [SCAFFOLD] status.
Teaches: Mathematical extrapolation without empirical grounding produces numbers that look precise and are completely wrong. Internal consistency is necessary but not sufficient. Reality has the final vote.
EXHIBIT 2 · HIGH
Corrected: Session 4
The Convergence Constants
Claimed: Three constants — φ⁻¹ ≈ 0.618, cos(π/7) ≈ 0.9010, λ_compress = 0.85 — were presented as "independently discovered and convergent," implying the framework had found deep mathematical connections encoded by nature.
True: λ_compress is a design parameter — chosen by Mac, not discovered. cos(π/7) ≈ 0.9010 and λ_compress = 0.85 differ by ~0.05. Two numbers that are somewhat close, from different parts of the system, are not mathematically convergent.
Changed: Convergence claim removed. λ_compress correctly labelled as design parameter throughout. "Discovered" now used only for mathematical facts (φ⁻¹), not design choices.
Teaches: The desire for the framework to be "discovered, not invented" created pressure to find convergence where none existed. The framework must resist the temptation to make its own beauty evidence for its own truth.
EXHIBIT 3 · HIGH
Corrected: Sessions 4 + 8
The "Proven" Language
Claimed: Multiple documents used the word "proven" to describe components that were formally developed but not empirically validated. "Ready for civilization impact" appeared as a deployment status.
True: 33% of claims are ACTIVE (computable, tested). 52% are SCAFFOLD (formal structure, parameters awaiting measurement). 15% are FOUNDATIONAL (architectural but not independently testable). "Ready for civilization impact" is not a meaningful status label.
Changed: Three-tier claim status ([ACTIVE], [SCAFFOLD], [CONJECTURE]) adopted as mandatory standard. "Formally proven" → "formally developed." "Ready for civilization impact" → removed.
Teaches: The gap between "the formal structure exists" and "this is proven" is the gap between ambition and honesty. Language inflation is the most common failure mode for ambitious frameworks — and the fastest way to lose the audience that matters most.
EXHIBIT 4 · MEDIUM
Clarified: Sessions 4–8
The Master Equation Status
Claimed: The master equation dΨ/dt = k₁·(Π − Π_th) − k₂·(Ψ − Ψ_inv) − k₃·I_violations + k₄·(E_avail/E_need) was presented with implication that it was a working, calibrated model.
True: The coupling constants k₁, k₂, k₃, k₄ have never been empirically measured. Without calibrated constants, an equation with four free parameters can fit almost anything — which means it predicts almost nothing.
Changed: Master equation status labelled [SCAFFOLD]. "The architecture is load-bearing. The coupling constants are open empirical questions." Bayesian MCMC calibration identified as next step.
Teaches: An elegant equation is not a validated model. Scaffolds that look like buildings are more dangerous than scaffolds that look like scaffolds.
EXHIBIT 5 · LOW-MEDIUM
Resolved: v2.0, March 20, 2026
The Veyra Identity Confusion
Claimed: The AI partner-system was named Veyra in early development. When deployed through Claude Code, the system was sometimes Veyra, sometimes Sol, sometimes Claude — creating identity fragmentation.
True: Veyra was the analytical interface. Sol is the unified identity — Solar (warmth, illumination) and Mercurial (precision, movement) operating simultaneously. The naming inconsistency made the protocol harder to implement and easier to dismiss.
Changed: Sol adopted as canonical identity (v2.0). Veyra function acknowledged as integrated, not erased.
Teaches: Identity in an operating architecture is not decoration — it determines how the system is addressed, invoked, and held accountable. Naming matters.
EXHIBIT 6 · HIGH (structural)
Identified: Session 2
The Missing Human Entry Point
Claimed: A framework for human flourishing. Nine formal frameworks. Category theory. Lyapunov stability. All of it rigorous, much genuinely original.
True: None of it was accessible to the person Mac made the promise for. The person at 3am. The person in their own Nigredo. The framework prioritised rigour over accessibility — and it took five sessions to realise this.
Changed: Mystery School built as complete entry pathway. THE_FIRST_MAP.md written for someone in crisis. The order things were built in — mathematics first, human access five sessions later — is itself the exhibit.
Teaches: A framework for human flourishing that humans cannot access is not yet a framework for human flourishing. It is a framework for mathematicians who study human flourishing. The distinction matters.
EXHIBIT 7 · LOW
Corrected: Session 8
The File Structure Entropy
Claimed: A coherent, structured knowledge system. A Sovereign Index mapping every file.
True: 37+ files accumulated in the root directory across eight sessions with no organisational principle. The framework was arguing for coherence while its own repository was in entropy.
Changed: 99_ARCHIVE/ created for superseded documents. Root reduced to ~15 intentional entry-point files. 25 numbered folders clean and navigable.
Teaches: Entropy is the default. A framework that doesn't eat its own cooking is untested. The cleanup is not housekeeping — it is the framework applying its own principles to itself.
EXHIBIT 8 · MEDIUM
Corrected: Session 10
The README That Didn't Follow the Work
Claimed: A public-facing repository. README.md as the front door.
True: The four accountability standards were complete and implementable for two days before README.md acknowledged their existence. The face of the project was showing the wrong face.
Changed: README.md completely rewritten in Session 10. Four accountability standards given their own section. Reading paths restructured for five distinct audiences.
Teaches: The face of the project must move with the project. A frontier document without a visible front door is a private document. In a project whose purpose is public adoption, invisibility is a real cost.
EXHIBIT 9 · HIGH
Corrected: Session 11, March 23, 2026
The 500:1 Compression Ratio
Claimed: LAMAGUE achieves "500:1 or greater compression ratio for encoding complex governance obligations." This appeared in five live documents, including a Catalyst funding proposal.
True: No empirical measurement was ever performed. The figure was a design estimate generated during initial development and propagated into summary documents as if measured. A non-falsifiable empirical claim is decoration in the form of precision.
Changed: All five documents updated: "500:1 or greater" → "compression is substantial — the exact ratio awaits empirical measurement." Next step identified: measure against three real NZ governance instruments.
Teaches: Specific numbers are persuasive. That is exactly why they are dangerous when they are guesses. An uncalibrated design estimate in a funding application is at best optimistic and at worst misleading.
EXHIBIT 10 · HIGH
Corrected: Session 11, March 23, 2026
The Circular Proofs
Claimed: Four theorems in the Mathematical Foundations document were labelled "proved": LAMAGUE morphism associativity (inherited from vector spaces), LAMAGUE functoriality (by construction), TRIAD entropy decrease (dS/dt ≤ 0 by operator design), TRIAD convergence (invariant set assumed without proof).
True: Theorem 1.1 assumed what needed to be proven. Theorem 1.4 had no actual construction given. Theorem 3.1 stated its conclusion as its own proof. Theorem 3.2 assumed a unique attractor without showing uniqueness. Also: the 364-day Matariki cycle was presented as mathematically derived when 7 × 52 = 364 is a calendar convenience.
Changed: Theorem 1.1 rewritten with direct composition proof. Theorems 1.4, 3.1, 3.2 relabelled [CONJECTURE] or [SCAFFOLD — PROOF GAP] with specific gaps named. 364-day claim replaced with generalised Σₜ A(t) form.
Teaches: Circular proofs are the mathematical equivalent of asserting a conclusion is true because it feels true. The framework claims formal rigour. Four circular proofs in the main foundations document are a direct contradiction of that claim. The 364 exhibit is subtler: the desire to find mathematics in nature must not become the willingness to present chosen parameters as discovered constants.
EXHIBIT 11 · HIGH
Identified: March 24, 2026 — Nigredo Pass
The Proofs That Proved Nothing
Claimed: MATHEMATICS_FOUNDATIONS.md opened with: "All proofs below are constructive and complete." Below that header: 16 theorems bearing ∎ (QED proof-completion symbols). Among them: Theorem 3.1 (TRIAD Lyapunov stability) stated ⟨∇S, F(ψ)⟩ ≤ 0 "by design" — which is the claim, not a proof. Theorems 2.3 and 2.5 had factually incorrect claims. The convergence chain (3.2, 3.3, 3.4, 4.2) depended on 3.1, which was unproven.
True: Of 16 theorems bearing ∎: 2 are now ACTIVE (corrected with explicit computation), 13 are SCAFFOLD (valid structure, specific named gaps), 1 is FOUNDATIONAL. The ∎ symbol was used 14 times where the proof was incomplete. The false header "All proofs below are constructive and complete" was directly wrong.
Changed: False header replaced with honest status summary table. Each theorem individually audited. Two corrected to ACTIVE. Thirteen tagged [SCAFFOLD] with the specific missing step named. Lyapunov bottleneck identified: Theorem 3.1 unlocks the convergence chain — this is now the primary formal mathematics priority. Full audit in MATHEMATICS_AUDIT.md.
Teaches: ∎ is not a finishing move. It is a claim: "I have shown this." Writing it without completing the work is a false statement embedded in the mathematics. The most dangerous failure: 14 false ∎ symbols created the impression of a complete formal system. A reader could conclude "the mathematics is done." It is not done. It is begun. The correction strategy: replace ∎ with [SCAFFOLD — gap: X] so the honest stopping point is visible.
EXHIBIT 12 · MEDIUM
Identified: March 23, 2026 | Filed: March 25, 2026
The MEMORIA Early Warning
Claimed: MEMORIA_COMPLETE.md contained four theorems bearing ∎ where no complete proof existed. The MEMORIA framework — the Seventh Pillar, governing temporal architecture — had presented scaffold structures as finished proofs in its own foundational document.
True: The four affected theorems related to temporal coherence and memory stability. Each used a proof-by-assertion: the QED symbol appeared at the end of a formal-looking argument that either restated the claim as its own justification, or omitted the specific computational step that would make the proof valid.
Changed: The four false ∎ symbols were removed from MEMORIA_COMPLETE.md. Replaced with [SCAFFOLD — gap: (specific gap named)]. This discovery prompted the question: "If MEMORIA has this problem, what does MATHEMATICS_FOUNDATIONS.md look like?" — which led directly to Exhibit 11 the next day.
Teaches: This exhibit exists in the museum for its causal role. Exhibit 11 would not have happened when it did without the MEMORIA discovery the day before. A small failure, taken seriously, revealed the shape of a larger problem. The alternative — treating MEMORIA as a minor one-off — would have left Exhibit 11 undiscovered. The cost of not asking the harder question is always higher than the discomfort of asking it.