Four Ways to Manufacture a Coordination Finding
A Control Battery for Mixed-Motive MARL
The group's research leads with multi-agent AI — coordination and its measurement pathologies, authorization and standing for artificial systems, and the governance of AI crises — alongside systemic risk and market microstructure, and the philosophy underneath both. One question throughout: how complex adaptive systems hold together when their parts want different things.
This page is the record: what is finished, what is with reviewers, what stalled, and what was abandoned. Software appears beside the paper it came out of. Work that is open rather than finished lives on Projects.
What the group is actively building on. The consent-friction formalism and the replicator-optimization mechanism are the theoretical core, and the multi-agent reinforcement-learning work that tests them is ongoing — a 3,750-run factorial built to break the framework did exactly that, and the corrected law and the post-mortem are both public.
A Control Battery for Mixed-Motive MARL
Restorative Voice Under Enforced Restrictions
Machine-checked proofs for results across the lab's papers
Through peer review and scheduled for publication.
Student-t GJR-GARCH with exogenous regressors in the variance equation
pip install gjr-garch-x
Submitted and with reviewers. Venues are named where a decision is pending.
Public, citable, and not currently under submission.
Operationalizing Standing Through Consent-Friction Dynamics
An Interpretable Crypto-Native Stress Composite for Retrospective Systemic-Risk Discrimination
A composite early-warning index for systemic risk at the DeFi–TradFi boundary
Hedging as the Fourth Money Laundering Stage
A Critique of Primitive Identity from ZFC to Homotopy Type Theory
Real results that do not yet carry the evidence their headline claims. Each one names what is missing, and several are open to collaborators.
A Computational Framework for Developmental Psychology
Why Training Data Conventions Persist After Safety Removal
Lines the group is no longer pursuing. They stay public because the arguments and data may be useful to someone else, and because a lab that only shows its live work is not showing much.
A Detection-Theoretic Derivation of the Born Rule
Full PDFs and Google Scholar metadata on each paper page. Also indexed at farzulla.org.