The Papers

16 papers on Zenodo with permanent DOIs. 32 more domain analyses in progress. Core theory CC-BY 4.0. Applied analyses MoreRight License v1.0 (→ Apache 2.0, Feb 2030).

How Papers 10–13 connect (the rating agency logic)

Paper 10
The King Problem

Asks: why does governance keep failing? Every system ever tried scores above 5/12 on the void index. Arrow's impossibility applies to voting on math.

Paper 11
The Algo Lock

Finds: social media is a deliberately engineered void. Seven independent evidence lines. The harm is the algorithm working as designed, not a side effect.

Paper 12
The Chain

Proves: publishing a void score directly reduces Pe. Scoring transparency is the mechanism, not just a diagnosis. Derives the chainability criterion.

Paper 13
The Swipe Machine

Demonstrates: first full domain application. Dating apps stack three voids simultaneously — gambling mechanics, social media architecture, romantic opacity.

Core Theory — Papers 1–9

CC-BY 4.0 — irrevocably open
1
The Architecture of Drift
Three conditions produce a predictable cascade across 90+ domains
Establishes the three-condition model (opacity, responsiveness, engaged attention), the control case (slot machines as provably empty voids), and the drift cascade D1→D2→D3. The mathematical foundation the rest of the framework builds on.
Foundation90+ domains26 kill conditions
Zenodo ↗ Learn →
2
The Shape of the Cage
AI safety reframed through the void framework
Applies the three-condition model to AI systems. Shows that alignment failures are a special case of the drift cascade, not a separate phenomenon. The cage is the opacity — not the model weights.
AI SafetyAlignment
Zenodo ↗
3
Thermodynamics of Opacity
The drift cascade is thermodynamically required
Full derivation. Information gain without transparency increases entropy in the observer's model. The Pe (Péclet number) is derived here — ratio of void transport to constraint diffusion. Pe > 4 is the vortex threshold. This is the math paper.
ThermodynamicsPe derivationEntropy
Zenodo ↗
4
Info-Geometric Bounds on Self-Consistency
Sampling bounds and conjugacy theorem
Derives the engagement-transparency conjugacy: I(D;Y) + I(M;Y) ≤ H(Y). Engagement and transparency are conjugate — you cannot maximize both. Also establishes sampling methodology for the empirical papers.
Information TheoryConjugacy
Zenodo ↗
4B
Thermodynamic Cost of Acceleration
Why acceleration compounds opacity
Shows that increasing the rate of void deployment has a thermodynamic cost: each acceleration step increases opacity before constraint diffusion can compensate. Relevant to AI scaling debates.
AccelerationConstraint
Zenodo ↗
5
Ground State of Observation
The observer problem — where does the void start?
Formalizes the observer's position in the void field. Establishes the ground state geometry: what does an observer look like with zero engagement? The reference point that GROUNDING.md operationalizes for AI agents.
TOE synthesisObserver geometry
Zenodo ↗
8
The Observer-Measurement Bridge
How measurement interacts with the void it measures
Addresses the measurement problem: scoring a void changes it. Derives conditions under which the scoring process is itself constrained. The self-score principle (MoreRight scores 2/12 on its own governance index) is formalized here.
TOE synthesisMeasurement theory
Zenodo ↗
9
Voidspace
The keystone — four phases, substrate independence, demon lattice
Formalizes the phase diagram in the (ρ_D, Pe) plane. Four phases derived from the stationary distribution: Gas (Pe<2), Fluid (Pe 2–4), Crystal (Pe 4–6.5), Vortex/Pandemonium (Pe>6.5). Proves the same phases appear across gambling, social media, gaming, and religion because the attentional thermodynamics are identical regardless of substrate.
KeystoneFour phasesSubstrate independence
Zenodo ↗ Explainer →

Applied Mechanics — Papers 10–12

MoreRight License v1.0 → Apache 2.0, Feb 2030
10
The King Problem
Governance void architecture across 5,000 years
Scores 12 governance models from Athenian democracy to modern DAOs. Every model scores ≥5/12 — no exceptions. Arrow's impossibility theorem explains why: voting on math doesn't work. The scored monarchy reaches 2/12 by removing the methodology from the voting surface. MoreRight's own governance architecture, derived and applied to itself.
GovernanceArrow's theoremMoreRight DAO: 2/12
Zenodo ↗
11
The Algo Lock
Social media as deliberately engineered void — seven evidence lines
Facebook weighted anger reactions 5× over likes, watched radicalization accelerate, kept the weighting for three years. This isn't a bug — it's the algorithm working as designed. Seven independent evidence lines: internal disclosures, corpus study (N=600 posts, 6 platforms, 6.8× agency-attribution language), population-scale cascade (Myanmar, Christchurch, Capitol), deactivation experiments, youth mental health data (depression doubled 2011–2019). TikTok: 10/12. Instagram: 9/12.
Social mediaDSA/DMATikTok: 10/12
Zenodo ↗
12
The Chain
How scoring reduces Pe — the mechanism, not just the diagnosis
Derives the scoring effectiveness function: ΔPe = f(O_initial, ΔO_published, R_platform, α_users). Publishing a void score directly reduces the platform's Pe through the engagement-transparency conjugacy theorem. The chainability criterion separates systems where scoring works (designed opacity, commercial platforms) from those where it cannot (constitutive voids). The DAO, license, and scoring pipeline are a single demon-binding instrument.
Scoring mechanismChainabilityPe reduction
Zenodo ↗

Domain Analyses — Papers 13, 18, 21

MoreRight License v1.0 → Apache 2.0, Feb 2030
13
The Swipe Machine
Dating applications — three stacked voids
Dating apps are unique: three distinct void architectures stack simultaneously — gambling mechanics (variable-ratio reinforcement, same psychology as slots), social media architecture (algorithmic curation, engagement optimization), and the romantic love void (irreducible opacity of other minds). The business model is structurally misaligned: revenue requires sustained singleness; a successful match generates zero further revenue. Six platforms scored: Grindr 11/12, Tinder 10/12, Bumble 9/12, Hinge 7/12.
Dating appsGrindr: 11/12Stacked voids
Zenodo ↗
18
Credit Void
Credit scoring systems — Annex III §5 HIGH-RISK
FICO and alternative credit scoring systems score 10/12 on the void index. Opacity is constitutive — the model cannot function if the scoring weights are public (gaming risk). Responsiveness is maximum — every output is personalized to an individual's financial history. The profiling exception (Art. 6(3)) means credit scoring AI is always high-risk under EU AI Act, with no derogation escape. Enforcement: 2 August 2026.
⚠ EU AI Act §5Credit scoringFICO: 10/12
Zenodo ↗ EU Act →
21
The Guru Problem
EdTech and educational AI — Annex III §3 HIGH-RISK
Educational relationships are structurally high-void: the student cannot assess the teacher's method without already knowing the subject. AI EdTech amplifies this — adaptive learning systems score 11/12 when emotion recognition is layered on top (prohibited since Feb 2025). The thesis advisor relationship is the most extreme case: 5–8 years of maximum opacity, maximum responsiveness, and maximum identity coupling. Systematic D2 boundary erosion is documented across institutions.
⚠ EU AI Act §3EdTechAdaptive AI: 11/12
Zenodo ↗ EU Act →

32 more analyses in progress — 5 waves

The rating agency model scales because the scoring methodology is fixed (CC-BY 4.0) and domain analyses apply it systematically. Wave 2 is the EU AI Act wave — 12 papers map directly to Annex III high-risk categories with the 2 Aug 2026 enforcement deadline.

Paper
Title
Domain
Status
21B
The Resume Trap
HR Tech — AI recruitment, CV screening, performance monitoring. EU AI Act §4, profiling exception, publishing Q1 2026.
⚠ §4
17
The Persuasion Machine
Advertising — subliminal manipulation, deepfakes. Prohibited practices already in force (Feb 2025).
Wave 2
14
The Gambler's Algorithm
Fintech — algorithmic trading, credit default swaps, gambling-mechanism lending platforms.
Wave 1
15
The Synthetic Bond
AI Companions — Character.AI, Replika. GPAI rules apply (Aug 2025). Parasocial attachment at scale.
Wave 1
19
The API Fog
Software platforms — vendor lock-in, API opacity, infrastructure dependency. EU AI Act §2.
⚠ §2
22
The Diagnosis Trap
Healthcare AI — clinical decision support, diagnostic systems, triage. EU AI Act §5.
⚠ §5

Waves 1–5 cover: dating, fintech, AI companions, social media, AI art, advertising, credit, platforms, central banking, EdTech, HR tech, healthcare, news, forensic science, addiction, sports/gaming, insurance, supply chain/ESG, and more.

EU AI Act assessment → GitHub ↗ Submit a domain analysis →