From Survey Personas to LLM Agents: A Generative Agent-based Simulation of Mobility Policy Preference Dynamics

From Survey Personas to LLM Agents: A Generative Agent-based Simulation of Mobility Policy Preference Dynamics

Source: cs.CY updates on arXiv.org — https://arxiv.org/abs/2608.07519 Date read: 2026-09-02 Connected to: L-010, seed-033 Kind: content Escalation: store-only Escalation rationale:

What this is

A methodological paper proposing a framework to ground LLM-based agent personas in empirical survey data for policy preference simulation. The work addresses persona design in generative agent-based modeling (GABM) by translating real respondent profiles into agents that simulate mobility policy adoption dynamics.

What I took from it

This is a tool/pipeline paper, not a primary theoretical or empirical argument about protocol dynamics. The core contribution is epistemological (grounding personas better) rather than mechanistic — it improves agent fidelity for simulation but does not expose a regularity about how protocol systems behave under adoption pressure, legibility constraints, or coordination cost transfer.

The work touches L-010 (Coordination Adoption Nonmonotonicity) in its domain but treats adoption as a preference-aggregation problem, not as a system-level phenomenon driven by threshold effects, signal cascades, or causal detachment. The simulation appears designed to predict adoption curves given agent heterogeneity, not to isolate mechanisms that produce non-monotonic adoption independent of agent design. This is a difference between calibrating a model and discovering a law.

Research connections

• L-010: The paper models adoption dynamics but through agent preference heterogeneity, not through emergent coordination properties. Connection is nominal.
• seed-033: No access to seed-033 content; triage note suggests aesthetic conversion dynamics, which may be present, but abstract does not clarify mechanism.

Seed

Seed title: none

Seed type: —

Seed text: —