Physics, Economics, Hurricanes and Mistakes
The context was a discussion of whether global warming increased the frequency of hurricanes. Mike Huben pointed to an argument by a friend of his: that hurricanes were more common at times when sea water temperatures were higher. This was offered as evidence that global warming would increase hurricane frequency, in response to opinions by hurricane experts that there was so far no reason to think it did.
Consider first the physics. A hurricane is a heat engine; it converts thermal energy into work in the form of swirling winds. In principle one could use windmills to turn the hurricane into, say, electrical power, although I doubt it is a practical project. A heat engine that simply turns heat into work is a perpetual motion machine of the second kind, impossible because it violates the second law of thermodynamics. Actual heat engines take heat from a hot source, convert some of it into work, and dump some of it into a colder sink. They run not off the absolute temperature of the source but off the temperature difference between source and sink.
The ocean is a thermal buffer. Because its heat capacity is large it averages out atmospheric temperature changes; that's why San Jose, where I live, has a milder climate than Davis, where my sister lives. Seasons when the ocean is unusually warm are also times when the temperature difference between sea and air is unusually high, which fits the observation that they are when hurricanes are likely to be generated.
Global warming, however, as the second commenter pointed out, affects both air and sea. There is no reason to expect it to change the difference between air and sea temperature, so the evidence that warm seasons generate hurricanes is irrelevant to the question of whether global warming would.
Next consider the economics, where the argument is less rigorous but still suggestive. One of the most famous mistakes in mid-twentieth century economics was the Phillips Curve. A. W. Phillips observed, correctly, that higher rates of inflation historically correlated with lower rates of unemployment. Economists concluded that there existed a tradeoff, that if a country was willing to accept a somewhat higher rate of inflation it could get a somewhat lower rate of unemployment.
The actual relationship was not between inflation and unemployment but between unanticipated inflation and unemployment. For reasons I sketch in one chapter of my price theory text, when inflation is higher than expected unemployment is lower as a result. On average, high inflation correlates with higher than expected inflation, hence Phillips' statistical result. But if a country adopts a constant inflation rate of, say, 5%, pretty soon everyone anticipates a 5% inflation rate and unemployment goes back up to its long run level. Eventually, after observing the theoretical argument confirmed in the form of "stagflation," economists recognized the mistake; it is now generally agreed that the long run Phillips Curve is vertical, that anticipated inflation does not reduce unemployment.
The logic of the economic mistake and of the hurricane argument is identical. An effect that depends on a variable being higher or lower than its usual value is misidentified as an effect of its being absolutely high or low.
Incidentally, readers interested in the dispute over how large the costs of global warming are likely to be may want to take a look at the comments on the Blog "Backseat Driving," where I have been arguing with people. The argument started with a post quoting this blog and responding and went on to an interesting post about the ways in which our views are affected by our political bias. Brian, whose blog it is, seems like a nice fellow, somewhat more reasonable than his allies in the comment section, but none of them, so far, is willing to seriously consider the possibility that he might be wrong.