"We examine the performance and robustness of monetary policy rules when the central bank and the public have imperfect knowledge of the economy and continuously update their estimates of model parameters. We find that versions of the Taylor rule calibrated to perform well under rational expectations with perfect knowledge perform very poorly when agents are learning and the central bank faces uncertainty regarding natural rates. In contrast, difference rules, in which the change in the interest rate is determined by the inflation rate and the change in the unemployment rate, perform well when knowledge is both perfect and imperfect"--Federal Reserve Board web site.