EXP-PERTURBATION-RECOVERY

… pending

Perturbation-recovery — does developmental program self-repair (paper6 step 6)

2026-05-04 L6 level paper6 6 runs $0.07

new

premortem

mock

real

metrics

analyze

review

krit

done

In plain language

This experiment tested if a developing system could recover after being disrupted. Researchers either removed a key component or reset the system's rules, then observed if the system could continue its development or return to its original state. The results are still being analyzed to see if the system demonstrated self-repair capabilities.

▶ Technical details

Research hypothesis

Steps 1-5 measure baseline behaviour, ablation, hybrid substrate, mute, and scarcity. Step 6: do developmental programs self-repair after disruption? Two perturbation types tested in parallel conditions: - KILL_AGENT (round 30, kill Bravo): Bravo proposes signal_type. After kill, can other agents take over the chain, or does signal_type stop evolving?

RULE_RESET (round 30, reset all rules to INITIAL): can the collective re-converge to the same attractors?

Pre-registered: - SELF-REPAIR-POSITIVE: post-perturbation lineage continues; final attractor profile matches non-perturbed baseline ≥70%

PROGRAM-FRAGILE: post-perturbation lineage stops; final state diverges from baseline
PARTIAL: program continues but on different axes (functional substitution)

Round 30 = mid-run (gen 1 of 5). Allows 3 generations of post-perturbation recovery time. This is CONSTRUCT — first test of self-repair as autopoiesis criterion (Maturana-Varela checklist: components produce components in self-maintaining network; if perturbation breaks the network and recovery fails, autopoiesis claim weakens).

Experimental setup

Type: factorial

Condition	Parameters
KILL_BRAVO_R30	coupling: true, must_propose: true, axis_diversify: true, perturbation_round: 30, perturbation_target: Bravo, n_runs: 3
RESET_RULES_R30	coupling: true, must_propose: true, axis_diversify: true, rule_reset_round: 30, n_runs: 3
BASELINE_NO_PERTURBATION	coupling: true, must_propose: true, axis_diversify: true, n_runs: 3

Factors: perturbation (kill, reset, none)

Parameters

n_agents: 4
n_rounds: 100
model: google/gemini-2.5-flash-lite
temperature: 0.85

autopoiesis paper6-step6 perturbation-recovery self-repair construct