Guest post from Jim Roumasset:
Critics (e.g. this NY Times article via Standupeconomist blog) of the Superfreakonomics chapter on global cooling (here) say that the science is settled and that skeptics are preaching nonsense. Two reasons for skepticism are the claims, in the next breath, that the widespread mid-70s belief in global cooling was the result of "a couple of media stories" and that Katrina was caused by global warming. Dubner's reply to his critics is here: Global Warming in SuperFreakonomics: The Anatomy of a Smear.
In the interest of promoting critical thinking, Plimer's Heaven and Earth goes much deeper than Levitt and Dubner. You can also Google "Nature, not Human Activity, Rules the Climate" for a shorter critique of IPCC 2007. All three critiques raise questions about the methodology of estimating climate models. My favorite is the quote in Nature attributed to John von Neumann: “Give me four adjustable parameters and I can simulate an elephant. Give me one more and I can make his trunk wiggle.” A climate model should also be capable of passing "the fingerprint test." The same model that explains variation in surface temperature data should be able to explain temperature patterns in the troposphere and the different patterns in the Northern and Southern Hemispheres. It is also instructive to read the critique of the now defunct hockey-stick analysis (from IPCC 2001) and the failure to use correct time-series methods in estimation of confidence bounds around mean surface-temperature records from tree rings and ice cores (Aufhammer, Yoo, and Wright, "Specification and Estimation of the Transfer Function in Paleoclimatic Reconstructions").
Note: there has been some slight editing since the original post.