For about 25 years before the COVID-19 pandemic, inflation was very low and stable in most advanced economies. A little noticed dark side of this impressive achievement is that unemployment rates were almost always higher than needed to keep inflation low. This widespread and persistent policy error arose because of a major flaw in standard macroeconomic models—the use of a linear Phillips curve. This flaw would have been far less costly if central banks had not chosen such a low target for inflation. This paper thus adds to the arguments in favor of a moderately higher inflation target. Even without a higher target, central banks need to use a broader range of economic models and should verify their estimates of the natural rate of unemployment by running the economy hot from time to time in order to see nascent inflationary pressure before throttling back.