EXP-RULE-PROPOSAL-PILOT

… pending

Rule proposal pilot — minimal autopoietic LLM collective (step 1 of paper6)

2026-05-03 L6 level paper6 2 runs $0.01

new

premortem

mock

real

metrics

analyze

review

krit

done

In plain language

This experiment tested if a small group of AI agents could create and change their own interaction rules over time. The agents were designed to propose changes to these rules, vote on them, and then use the accepted rules to guide their future actions and proposals. The results are still being analyzed to see if the agents successfully developed new rules or if the system became too simple.

▶ Technical details

Research hypothesis

Minimal autopoietic loop: agents propose coupling-rule modifications, vote, activate them. Rules govern how agents interact. Interactions produce next proposals. Loop: agents → rules → coupling → agents. Pre-registered: - Positive: ≥3 rules differ from initial state by round 40; ≥1 proposal accepted via voting; rule history shows non-trivial trajectory (not all proposals from same agent, mix of accepted and rejected).

Negative: 0-1 rule changes; all proposals same axis; voting always unanimous (rubber-stamp) or always rejected (deadlock).

Either outcome substantively informs paper6: does minimal autopoiesis work in LLM substrate, or does the system trivialize?

Experimental setup

Type: simple

Condition	Parameters
AUTOPOIESIS_PILOT	coupling: true, n_runs: 2

Factors: coupling (true)

Parameters

n_agents: 4
n_rounds: 40
model: google/gemini-2.5-flash-lite
temperature: 0.85

autopoiesis self-modifying-rules paper6-foundational rule-space-evolution