On a recent article at ATTP (wherein it is implied that Michael Tobis should not be famous, hrmph) Willard points to a very nicely written 2011 article by John N-G,
Roger Pielke Jr.’s Inkblot which discusses the oft-confused ideas about whether climatology makes testable predictions.
But science doesn’t work by making predictions about future events, for the most part; it makes predictions about observable aspects of the world, things detectable in the present. The amount of trust scientists place in climate models, for example, depends on their ability to simulate relevant aspects of the past and present world. The amount of trust the public places in climate science should depend on the weight of evidence in the past and present world, which is enormous.
Sorry for the late news, but this one is a keeper. Better late than never, I guess.