Ideas I have for potential PhD theses.
Self-Organizing Symbolic Systems for Emergent Computation: Build a mathematical and computational framework for symbolic systems (like CRDTs + Lambda Calculus + Git) that evolve higher-level programs through local interactions.
Decentralized Inference in Dynamic Networks: Beyond CRDTs: Formalize and implement new models for consistency, belief revision, and inference over dynamic, failure-prone distributed networks — blending CRDTs, epistemic logic, and adaptive synchronization.
Computational Ontologies for Emergent Meaning in Multi-Agent Systems: How symbolic meaning and structure can emerge without explicit central control in systems of communicating agents - combining philosophy of language, cognitive science, and multi-agent systems.
Phase Transitions in Artificial Self-Organizing Systems: Study criticality, phase transitions, and emergence of computation in designed systems (e.g., gossip meshes, agent-based simulations), connecting real computation with edge-of-chaos dynamics.
Energy-Based Learning and Embodied Computation Beyond Deep Learning: New architectures for embodied, predictive AI systems inspired by energy-based models, dynamical systems, and ecological psychology (Gibson, Varela).
Rust-Based Frameworks for Dynamically Evolving Distributed Systems: Develop safe, composable primitives for runtime-evolving distributed graphs (services, agents, beliefs), pushing Rust and WebAssembly into adaptive computation territory.
Modeling the Evolution of Computational Epistemologies: Formal models for how agents (human, machine, hybrid) build, revise, and transmit systems of knowledge over time - incorporating uncertainty, revision, and metaphor as first-class citizens.
Emergent Governance in Mesh Networks: How distributed clusters (human or machine) can create governance structures and trust networks dynamically, without centralized control, and resilient to churn.
Time as a Computational Primitive in Distributed Systems: Rethink "time" in distributed computation: drift, divergence, and coherence without a global clock - inspired by relativity, biological systems, and event-driven computation.
Symbolic-Manifold Hybrid Representations for Learning Systems: Blend discrete (symbolic) and continuous (manifold) models for AI systems that reason and intuit — going beyond current neuro-symbolic models into emergent hybrid forms.