Nate Silver has published a book, The Signal and the Noise: Why So Many Predictions Fail -- but Some Don't, that I can heartily recommend to anyone who wants to evaluate the many predictions we encounter around us and that we make ourselves. It's an argument for utilizing Bayesian probability, an application of mathematical logic, in how we think about the world; if that sounds forbidding, just let me say that Silver's book is not difficult at all. He's a sharp observer of multiple subcultures, including sports, gambling, weather forecasting, economics, and earthquake science. In short, this book is fun and I thought a convincing argument in favor of a mode of prediction that might improve our understandings.
In particular Silver puts the arguments about the reality of climate change in a context I found slightly different than I'd understood. There are people for whom not believing in the climate conclusions of mainstream science is a good gig (sponsored usually by the fossil fuel industry) or who thrive on pure contrarianism, so it possible to say that "not all scientists" agree that global warming is real. But it is possible to discern which climate assertions win wide agreement and which are open to appropriate scientific questions.
It is useful to the mere interested observer of scientific argument to have to parameters of consensus laid out so clearly.
The fact that some issues still need more data points and more sophisticated models in order to generate predictions that command overwhelming assent would not be a problem if climate change were merely a puzzle for scientists. But the scientific mode of knowledge has wandered into a political minefield where its conventions don't fit social needs.
Responsible prediction is hard; Nate Silver at least makes its contours more accessible in this book.