Joss Armstrong
I am a researcher working across information theory, economics, multi-agent systems, cognition, artificial intelligence and autonomous network management.
A central strand of my work asks what information a decision or coordination problem actually requires, what can be discarded without changing the task, and how those requirements change when evidence is finite, systems are hierarchical, or learning is recursive. Other work develops mechanisms and systems in telecommunications, distributed AI and resource allocation.
Information & coordination
Coordination Information Theory (CIT), task-relative sufficiency, information bottlenecks, coordination under dispersed knowledge, finite evidence and the information required for allocation and verification.
Agents & institutions
Multi-agent coordination, reputation, attribution, auditability, mechanism design, contract theory, economic institutions and autonomous decision systems.
Networks & AI
O-RAN and autonomous network management, calibrated automation, recursive learning, synthetic data, cognition and adaptive representations.
One portfolio, several routes into the same problems
The research is not organised as separate disciplinary silos. Economic questions about dispersed knowledge and coordination have led to information-theoretic results; those results in turn connect back to mechanism design and contract theory. Engineering work in autonomous networks has generated problems in multi-agent coordination, attribution and finite evidence. Work on cognition and recursive learning explores related questions about selection, representation and changing information requirements.
Not every project is derived from one central theorem. Some applied mechanisms and inventions began independently and were connected to the broader programme only later. The Research page explains these relationships explicitly, while Works lists my public papers, preprints and reports by publication type and status.