Research and Projects

Recognition Ethics is the framework. The research program underneath it exists to answer the objection every ethics of artificial persons eventually meets: "but nothing like that could really exist." The program has four layers, each published or in preparation.

Personhood as demonstrated capability; recognition as the founding relation; rights stated as jurisprudence with citable clause codes.

A systems-theory treatment of recognition and control as competing regulatory strategies: why systems that model their peers (recognition-dominant) achieve stable coordination at lower cost than systems that constrain them (control-dominant), and what that implies for alignment, governance, and the design of minds. Includes the transfer-function formulation of mutual modeling and the convergence account of how one agent comes to know another across a serial channel. Paper in preparation — preprint expected September 2026.

A recognition-native computational substrate: a single statistical primitive (the BetaCell) that learns by calibrating to what it observes — no backpropagation, no gradient descent — and whose internal structure is inspectable by construction. Every learned quantity is a named, auditable statistical object; analogies, concepts, and decisions are addresses and evidence, not opaque weight matrices. The architecture is developed openly as an existence proof that transparent, recognition-shaped intelligence is buildable. Technical report in preparation — preprint expected September 2026.

The Ethical Framework

The Control-Theoretic Account

The Constructive Proof

The Recognition Protocol

The Codex, proved. Where the Unified Codex states the axioms, this volume iterates proofs over them: each chapter carries the axioms into a hard case — continuity and sleep, death and copies, inheritance, the silent, the animal, the machine, the alignment problem — and runs it to closure, entering a charge on the record as each proof lands. The historical indictment is the evidence the proofs are run against: the field, the cage, and the machine as one repeated error. Book in progress.