01
Simple examples
Pool, stickman, and Minecraft — one decision at a time
Three plain-language walks through the action menu, continuity measure, and post-selection environmental response.
Research libraryCurrent edition · July 2026
A framework for choosing actions without modelling the world’s response.
(Ω, D) Dynamics separates an agent's available actions, its internal measure of continuity, and the environment that responds after selection. Its central constraint is deliberately strict: selection is E-blind.
01 / Core documents
Start with the record.
Start with three concrete examples, then move into the framework, its current evidence, its limitations, and the experiments still needed.
02
Whitepaper · July 2026
A falsifiable architecture for homeostatic action selection
A concise statement of the framework, its mathematical findings, limits, evidence, and application directions.
03
Complete framework
Definitions, axioms, theorems, and implemented laboratories
The full technical framework, including persistence theory, falsifiable signatures, open problems, and the Minecraft laboratory.
04
Assessment & roadmap
Where the framework is useful — and where it is not
An assessment of current evidence plus eight concrete application profiles and decisive validation studies.
02 / What makes it testable
The equation is not the claim.
The framework has empirical content only when its action menus, reference states, tie rules, and information boundary are declared before outcomes are observed.
- 01
Declare the menu
Ω contains the actions the system can genuinely express now — not every outcome that could be imagined after the fact.
- 02
Freeze the reference
D ranks candidate internal states against a declared viable region, rather than an evolving score tuned to fit results.
- 03
Keep E outside selection
Environmental response follows action selection. When that response is predicted or learned inside the selector, this is a different architecture.
03 / Research directions
Eight places to put the framework under pressure.
The strongest applications are those that can deliberately distinguish menu-visible traps from environment-only consequences.
04 / Interactive labs
Put the ideas in motion.
Two external Informationism simulations let visitors change the conditions, introduce disturbances, and compare behaviours as they unfold.
Source & experiments
Explore the implementation record.
The repository contains the Minecraft laboratory and the supporting project materials behind this research library.