Search papers, labs, and topics across Lattice.
This paper introduces a causal counterfactual framework for evaluating policy interventions within LLM-based social simulations, distinguishing between necessary and sufficient causation to address the distinct needs of moderators and platform designers. They formalize the mapping between these causal concepts and stakeholder needs, outlining how simulation design can support estimation under explicit assumptions. They argue that resulting estimates should be interpreted as simulator-conditional causal estimates, emphasizing the importance of simulator fidelity for policy relevance.
Simulations that look realistic aren't enough: to reliably test governance interventions, we need to move to causal simulations that can support policy changes.
LLM-based social simulations can generate believable community interactions, enabling ``policy wind tunnels''where governance interventions are tested before deployment. But believability is not causality. Claims like ``intervention $A$ reduces escalation''require causal semantics that current simulation work typically does not specify. We propose adopting the causal counterfactual framework, distinguishing \textit{necessary causation} (would the outcome have occurred without the intervention?) from \textit{sufficient causation} (does the intervention reliably produce the outcome?). This distinction maps onto different stakeholder needs: moderators diagnosing incidents require evidence about necessity, while platform designers choosing policies require evidence about sufficiency. We formalize this mapping, show how simulation design can support estimation under explicit assumptions, and argue that the resulting quantities should be interpreted as simulator-conditional causal estimates whose policy relevance depends on simulator fidelity. Establishing this framework now is essential: it helps define what adequate fidelity means and moves the field from simulations that look realistic toward simulations that can support policy changes.