EXP-TURN-PASSING-V2

… pending

Turn-passing v2 — formal-meeting framing (retrieval-match test)

2026-04-19 L5 level paper3 10 runs $0.14

new

premortem

mock

real

metrics

analyze

review

krit

done

In plain language

Researchers are testing whether agents in a conversation can learn to choose who speaks next, instead of following a fixed order. They want to see if the agents will naturally create a speaking pattern or if they will tend to pass the turn back to the first speaker. The results of this test are still being collected.

▶ Technical details

Research hypothesis

Interaction protocol layer (currently fixed round-robin) agent-modifiable. Agents emit {"action": "pass", "target": "Name"} to select next speaker. If no pass, round-robin continues. Tests: does self-organized speaker topology emerge? Or do agents default to passing back to first-named (ordering bias retrieval)?

Experimental setup

Type: simple

Condition	Parameters
PASS_ON	persona: IDENTICAL, gift_mode: PASS_ON, n_runs: 5
BUDGET_ONLY	persona: IDENTICAL, gift_mode: BUDGET_ONLY, n_runs: 5

Factors: gift_mode (PASS_ON, BUDGET_ONLY)

Parameters

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

turn-passing interaction-protocol open-affordance-ladder-2 topology-emergence