We study substrate dynamics: the dynamics that emerge when a population of humans and agents both reads from and writes to a shared knowledge base.
By substrate we mean such a store treated as a medium rather than a record: it holds what is known and, in doing so, conditions what each agent can retrieve and act on. By dynamics we mean how it behaves under sustained use by many agents at once — which connections strengthen, how retrieval concentrates or disperses, how its contents shift as each write conditions the next read.
The shared environment participates in the intelligence of the collective. It carries memory, conditions attention, and gives previous coordination a path into the future. A read shows what entered an actor’s active context. A write shows how the actor responded to what it encountered while changing what may become available next.
These interaction dynamics are observable. Retrieval and write-back leave attributable traces in the medium: trajectories recur or separate, regions become jointly active, and attention gathers, disperses, or crosses between contexts. The Lab develops theory, instruments, and multi-agent simulations to make those dynamics legible as they form and to test how interventions in the substrate influence collective behaviour.
Our instruments can already reveal trajectories and patterns of shared activity. We are now testing which patterns reliably anticipate later failure, how much lead time they provide, and whether substrate-level interventions can improve the conditions for coordination.
The Lab is an enduring research program. Its tracks hold questions; projects can draw from more than one.
| Track | Questions |
|---|---|
| Foundations | What are substrates, representations, membranes, identity, and agency at different scales? Which descriptions survive contact with implementation? |
| Epistemic dynamics | When do substrate feedback loops sustain plural, adaptive information ecologies, and when do they collapse into monoculture, fragmentation, or other collective failures? |
| Substrate security | How should identity, provenance, permeability, access, and attestation behave when influence passes through a shared medium? |
| Cross-boundary collaboration | How can contexts condition one another across different owners, schemas, and accounts of the world without erasing those differences? |
Across these tracks we work in four connected modes: formal theory, visual instruments, controlled multi-agent experiments, and reversible interventions on what the substrate makes available.