A PARADOX IN BRIAN ARTHUR’S SOLUTION TO HIS EL FAROL BAR PROBLEM

El Farol is a bar in Santa Fe, California, that people choose to visit only if they expect it to be uncrowded; otherwise, they prefer to stay home. How do people choose to go or stay home? Our study uncovers a paradox that emerges when applying Brian Arthur’s original “method of predictors” to address this problem. Arthur’s method was designed to prevent the convergence of expectations by introducing a diverse set of “predictors”. However, as agents increasingly rely on more accurate predictors, their expectations tend to synchronize. This explains the widely fluctuating attendance around the optimal average attendance previously noted in the literature. This paradoxical situation is not trivial, as it underscores the limitations of the method of predictors to address the El Farol bar problem as one of congestion of infrastructure. However, if we transition from an objective to a subjective performance metric, Arthur’s method does offer a viable solution to the problem.