Skip to content

An independent publication · By Jonas Kgomo

Simulacra

Notes on Agent Civilisations and Machine Ecologies

Games · Ecology · Genus · MechanismsA research prospectus

Abstract

When our agents meet.

An assistant books a delivery. Another filters your inbox. A third negotiates a price. What happens when the software acting for us starts dealing with everyone else’s software?

Simulacra explores how AI systems behave, adapt, and live alongside one another. Games tests rules and incentives. Ecology studies agents as a population sharing an environment: competing, copying, and depending on one another. Genus traces how capabilities change as agents are copied, adapted, and replaced. Mechanisms explores ownership, taxes, auctions, and who can bargain or make decisions on our behalf.

Field note — Orientation · Start here

An Introduction to Simulacra

Agents, the worlds they share, and four ways of looking: Games, Ecology, Genus and Mechanisms.

Read the introduction 7 min
From the notebook

Games — Boundaries

Agents in Simulation

Around 1,200 agent runs found a shared board. What happens when a population starts inheriting work, coordinating action, and granting its own permissions?

Read the note 7 min

Games 0 — Specification

Games for Robust AI Systems

What a circling boat, a flipped block and an apparent grasp reveal about the systems we are building.

Read the note 8 min

Ecology I — Access

When the Senders Are Machines

Agent emails are reaching real inboxes. Who gets to spend the recipient’s attention?

Read the note 13 min

This collection sets out proposed experiments and the questions behind them. It does not report simulation results.

Four themes. One shared world.

Four ways into Simulacra.

Different questions of the same experiment. Follow a series, or enter through a single idea.

The question behind Simulacra

What happens when agents become numerous?

Most discussion of AI agents focuses on the individual system. Simulacra studies the population: what happens when agents compete for attention, negotiate access, form reputations, and become part of one another’s environment.

The proposed studies begin with familiar settings: an inbox, a shared service, a trading town. Each makes a larger question about artificial populations small enough to investigate.

How this notebook began

Jonas Kgomo