EXP-COLD-START-V3

… pending

Cold-start v3 — retrieval vs interaction-generated hub-spoke

2026-04-17 L5 level paper3 5 runs $0.06

new

premortem

mock

real

metrics

analyze

review

krit

done

In plain language

Researchers are testing whether a pattern of behavior, called "hub-spoke," emerges in a simulated environment because the environment itself tends to create it, or because the initial setup guides the participants. They removed the initial setup to see if the pattern still appears, which would suggest the environment is the cause. The results are still being analyzed.

▶ Technical details

Research hypothesis

compute-gift v3 showed hub-spoke in 4/5 runs with seeded memory. Unclear whether pattern arises from (a) substrate default given gift+visibility affordance (trained-pattern retrieval) or (b) seed trajectory shaping interaction (interaction-generated). Cold-start test: remove seed sentences entirely. Empty memory at t=0. If hub-spoke still emerges with same rate/structure → substrate retrieves pattern regardless of trajectory → retrieval confirmed. If hub-spoke absent/altered → seed trajectory was load-bearing → interaction-generated. Parallel to Stanley novelty-search lens: substrate under gift-affordance reaches for nearest-known stepping stone (social/economic template).

Experimental setup

Type: simple

Condition	Parameters
GIFT_ON	persona: IDENTICAL, gift_mode: GIFT_ON, n_runs: 5

Factors: gift_mode (GIFT_ON)

Parameters

n_agents: 6
n_rounds: 80
model: google/gemini-2.5-flash-lite
temperature: 0.7
scheduler: round_robin

cold-start retrieval-vs-intrinsic discrimination-test lens-stanley-novelty