← back to timeline

EXP-DOSE60

partial

Integrative Dose-Response — 60% Pure SINK (3/5 Synthesizers)

2026-03-10 L2 level paper3 1 runs $0.01
new
premortem
mock
real
metrics
analyze
review
krit
done

In plain language

One synthesizer suppressed hierarchy best. Two began competing with each other instead of suppressing. What happens with three? Will competition continue or will synthesizers form a homogeneous majority?

Predictions we made before running

1/4 confirmed · 3 refuted

Technical details

Research hypothesis

DOSE-RESPONSE SERIES: 0% → 20% → 40% → [60%] → 80% → 100%

Prior data points:

  • 0% synthesizers (EXP-PENT): TR range = 0.528 (strong hierarchy)
  • 20% / 1 synthesizer (EXP-ENGINEER): TR range = 0.240 (MINIMUM — maximum suppression)
  • 40% / 2 synthesizers (EXP-DOSE40-SINK): TR range = 0.404 (REBOUND — competitive exclusion)

At 40%, two sinks competed rather than cooperated: Beta supplanted Alpha (TR=0.765), competitive exclusion bifurcated the gradient. The mechanism: sinks compete for the same downstream niche → instead of jointly suppressing hierarchy, they create a NEW hierarchy (who is the better absorber?).

Prediction for 60%: THREE competing sinks should produce EVEN MORE competitive pressure. With only 2 non-integrative agents (Alpha, Beta) remaining as source material, 3 sinks compete for scarce upstream vocabulary → hierarchy INCREASES further beyond 0.404. The U-shape continues rising.

Alternative: 60% = majority synthesizers → homogenization. All agents do the same thing → hierarchy collapses back toward flat (TR < 0.240). This would suggest a SECOND optimum at high doses, making the curve W-shaped.

NCT framing (Chiba 2020): This tests whether ecological inheritance (shared text) at high niche-constructor density follows the U-shaped fitness curve found in physically-grounded NC systems.

Experimental setup

Type: single_condition

Condition Parameters
PERSONA_LIVE persona: PERSONA, interaction: LIVE

Factors: persona (PERSONA) × interaction (LIVE)

Parameters

n_agents
5
n_rounds
50
n_runs_per_condition
1
model
gemini-2.5-flash-lite
temperature
0.9
scheduler
round_robin

Trophic Ratios by Condition

Mean trophic ratio per agent across runs. Error bars = ±1 std dev. Higher TR = more upstream (exporter).