KITE speeds human decision experiments with fewer model runs
KITE: Scaling Jev Population Experiments with Sparse Flagship Calibration
Multiagent SystemsComputers and Society
Summary
Studying how people behave using computer simulations can take a lot of time and computing power. The authors developed KITE, a method that runs detailed model calculations only on key situations and uses a simpler summary for many others. This approach lets them simulate large groups quickly while still measuring how close the model is to real people. Their tests showed that KITE reduces errors and runs millions of simulated steps very fast on normal laptops.
What this means in practice
- •For policy analysts: Evaluate the impact of policy interventions using fewer computationally expensive simulations without losing accuracy.
- •For social science software developers: Implement fast behavioral population simulators that maintain fidelity to expensive benchmark models by mixing sparse anchors with tabulated results.
Authors
Hengyu Li
Abstract
KITE queries a typed behavioral kernel once per unique state, then executes populations of any size from the table with event-keyed randomness and common random numbers. An expensive flagship model is reserved for sparse paired anchors that estimate intervention effects. Measured human-model discrepancy is propagated as shared error into every conclusion. Population-experiment cost thus scales with unique states and anchors, while uncertainty is governed by evidence about people rather than Monte Carlo noise. On Epstein experiments with 9,070 participants, anchors covering 1.7% of states reduced effect error by 41% (absolute MAE reduction 0.0125). On 37 held-out SocSci210 experiments, 0.5-1.5% anchor coverage raised captured decision gain from 0.27 to 0.39. The kernel passed content-fidelity criteria in all 15 new countries of a 16-country study. Shared discrepancy yielded retrospective coverage of 93% and 96% at nominal 80% and 90%, versus 29% and 36% from human sampling uncertainty alone. A million agents executed 20 tabulated steps in 0.9 seconds on a laptop. This architecture offers a route to screening candidate interventions before human trials, multi-country content audits, and uncertainty-aware policy comparison at the cost of a few thousand kernel calls with sparse flagship anchors. Property-specific evidence records connect each use to its validation scope, correction provenance, and uncertainty, making these applications auditable.