Welcome to Dooders, an open-source project researching the emergence of intelligent agents within a computational model of reality. Its mission is to advance the understanding of agency through continuous exploration and experimentation — studying how simple agents, placed under real selection pressure inside simulated worlds, develop increasingly capable and adaptive behavior.

Most of this work is led by Chris Mangum and lives in the open, with AgentFarm serving as the flagship experimental platform.

All research and development is in the early stages.

• Emergent agency & multi-agent systems — how populations of interacting agents discover niches, cooperate, and develop higher levels of causal control over their environment.
• Neuroevolution & reinforcement learning — evolving heritable hyperparameter genomes and decision policies, and probing when learning beats its own initialization.
• Complex adaptive systems — treating simulation as a microscope, measuring fitness, selection, and inheritance with paired seeds, confidence intervals, and robustness gates.

AgentFarm is a Python-first platform for agent-based simulation, reinforcement learning, and emergent-behavior research:

• Multi-agent simulations with configurable genomes, actions, and spatial indexing
• An experiment runner with SQLite-backed metrics and an analysis pipeline
• A decision stack spanning DQN, distillation, and neuroevolution experiments
• REST/WebSocket API and structured logging, all wrapped in a documented, reproducible workflow

Docs: dooders.github.io/AgentFarm

Selected results from the AgentFarm research devlog:

• The transferable-signal gate: do learned policies beat their own init? — paired held-out rollouts show a modest but robust early-age decision-quality signal worth inheriting.
• When every agent has a different goal — making the reward function itself a per-agent, heritable trait; un-curated objective diversity measurably lowers collective fitness across 20 paired seeds.
• Are we measuring at the wrong level? — re-scoring a 36-run inheritance A/B at the newborn level reveals small robust behavioral shifts, while the population-level null held.

• Research platform for multi-agent systems with AgentFarm.
• A modular, pluggable memory backend for AI agents with AgentMemory.
• Modeling a biological cell with Pyology.
• Exploring the Agent and Arena relationship with Dooders.

Foundational and exploratory projects that informed the current work:

• Representation and meaning preservation in agent memory with AgentMeaning.
• Physics-inspired universes with Fizicks.
• A temporal simulation engine with TimeBandit.
• Evolutionary algorithms with ContinuousEvolution.
• Autonomous and independent agents with AnarchyGraph.

• Devlog: build notes, design decisions, and experiment outcomes — dooders.github.io/AgentFarm/research/devlog
• Researcher: Chris Mangum — Independent Researcher

Any contributions are welcome! Whether it's bug fixes, new features, or improving documentation, your help is appreciated. Please refer to the CONTRIBUTING.md file in each project repository for guidelines on how to contribute. Or see the general version here

For any inquiries or further information, please contact at doodersai@gmail.com.

All projects under the Dooders organization are open-source. See the LICENSE file in each repository for the specific license and more details.