DragonTekna · Archive

Research Papers

Technical papers produced by the DragonWorx team — companions to the essays published here. Each paper represents the formal research layer that sits beneath an article.

Mathematical Science

Number Theory · Random Matrix Theory · July 2026

One Object, Two Faces: Spectral Duality Between Riemann Zeta Zeros and Prime Numbers

A computed essay demonstrating the bidirectional duality between the Riemann zeta zeros (rigid, GUE-correlated spectrum) and the primes (Poisson-like sequence). Figures computed from 2,000 zeros via mpmath and primes sieved to 10⁷. Includes the Goldston–Montgomery equivalence, number variance, and pair-correlation form factor — each result verified numerically.

Mathematical AI & Architecture

Automated Reasoning · AI Architecture · July 2026

Beyond Machine Persistence: Weaknesses of Single-Premise Proof Swarms and the Design of a Dual-Premise, Obstruction-Exchange Architecture

Nine structural weaknesses in the OpenAI Cycle Double Cover proof architecture, followed by a full specification of DPOX — a dual-premise system where two swarms search simultaneously for proof and counterexample, exchanging structured failure records until the boundary of the problem is charted.

Governance & Mechanism Design

Political Philosophy · Mechanism Design · June 2026

The Calibrated Preference–Forecast Mechanism: A Complete Working Dossier

A governance mechanism that separates what a population wants (egalitarian cardinal voting) from what will happen under each policy (forecasts weighted by calibration track record, not money). Includes Stage-0 theorems, Stage-1 agent-based simulation results, and a staged empirical verification program. The formal core of the anti-plutocracy claim: averaging stays bounded, accumulation concentrates.

Computational Biology

🧬
Computational Biology · AI · June 2026

The Mothersoul Architecture: Multi-Agent Simulation of C. elegans Neural Dynamics with Emergent Cultural Transmission

A proposal for a multi-agent simulation in which individual AI minds inhabit the 302-neuron connectome of C. elegans, live full simulated lifetimes, and return learned behaviors to a collective memory pool. Explores emergent culture, pharmacological modeling, and the architecture of death-and-return knowledge propagation.