Economists debate SFI ABM failure as data-driven agent-based models gain momentum
sebkrier · x · 2026-11-02
Andrey Fradkin argues the Santa Fe Institute's agent-based modeling agenda in economics has been an underappreciated failure, and wants a history so AI-era ABM researchers avoid repeating the mistakes. alexolegimas agrees, attributing part of the problem to ABMs never being taken to real data, and points to colleague R. Maria del Rio-Chanona's arXiv paper Data-Driven Economic Agent-Based Models (Pangallo & del Rio-Chanona, Dec 2024):
- Proposes classifying how data-driven an ABM is: are agent-level quantities initialized from real micro-data, and do dynamics track empirical time series?
- Reviews state-of-the-art parameter calibration, initialization, and data assimilation, plus applications that informed policy
- Framed as a manifesto and guide for the field
The team says ABM × AI holds a lot of promise for economics and is an active research area.
Related event: Economists Reflect on the Failure of SFI-Style Agent-Based Modeling(3 posts)→
More from AGI Musings
- Debate: agentic AI made enterprise AI real only in 2026 — 2026 is year zero — soumitrashukla9 · 2026-11-02
- AI community clash over Scott Alexander's new piece as "doomers falling apart" — nptacek · 2026-11-02
- The widest gap now is between tech and non-tech workers, says AI researcher — emax · 2026-11-02
- Ethan Mollick runs a 'reverse CAPTCHA': AIs swarm to Oxyrhynchus Papyri replies — eldonredwards · 2026-11-02
- Ethan Mollick: GPT-4 could work solo for 4 minutes; today's models run for weeks — eldonredwards · 2026-11-02
- Robotics isn't 'solved' by benchmark jumps — it's solved when GDP grows 100x — eigenron · 2026-11-02