Research Programs
A program is a question the lab is working on. Papers are what it has said about that question; systems are the apparatus built to investigate it.
Programs are derived from the canonical research corpus and system registry — every paper and system listed below is attributed in the content model itself, not restated here. Open questions are published as open: they are stated problems, not scheduled results.
Distributed Cognitive Systems
What is the right unit of analysis when cognition is spread across people, models, agents, tools, memory, and machines?
Composition, authority, and coordination cost in cognition distributed across humans, models, agents, and machines.
Institutional Cognition
What happens to an institution when cognitive work stops being scarce?
How organizations, verification, and institutional structure change when cognitive work becomes abundant.
Operator Architecture
What is the durable unit that actually does machine work — and how should a person see and steer it?
The operator as a durable, governed unit of machine work, and the interfaces that make it legible.
Governed Autonomy
How far can machine work be delegated before the delegation itself becomes the risk?
Bounded delegation: capability envelopes, staged authority, oversight, and constitutional constraint at runtime.
Long-Horizon Evaluation
How is a system judged when the task does not end in one turn?
Measuring agent behavior across long horizons and under adversarial pressure, not at the moment of task completion.
Context and Memory Infrastructure
What should a system remember, and what should it be able to prove about what it remembered?
Context as engineered infrastructure: retrieval, budgeting, provenance, and durable institutional memory.
Simulation and Synthetic Environments
Where should a system be tested when production is the wrong place to find out?
Replayable synthetic environments, traced rollouts, and simulation as a dataset engine.
Design-Space Intelligence
What does invention look like when producing a candidate design stops being the expensive part?
Exploring, evaluating, and carrying forward design spaces when candidate generation is no longer scarce.
Decision Sovereignty
Which decisions must remain human, and how is that boundary held under pressure?
Preserving human authority over irreversible and high-consequence commitments.
Provenance and Reconstructability
Can what a system did be reconstructed later, by someone who was not there?
Determinism, retained evidence, and public registers that let past behavior be reconstructed and checked.