After several years of waiting with baited breath, the National Research Council's rankings of doctoral programs in the US is here. In an effort to be true scientists, the rankings come with up to 61 categories which can be weighted in any way your heart desires. Do you want to see rankings based on average workspace per grad student and availability of academic ethics training? Then look no further.
Another interesting, helpful, and extremely complicated feature is that instead of producing a single set of rankings, they have created probability distribution functions for each schools ranking and then listed the 90% confidence interval. That means that in Astronomy and Astrophysics programs Harvard, Caltech, UC-Berkeley, and MIT are all ranked "1 - 9", with Arizona and Princeton ranked "1 - 10" and Johns Hopkins ranked "1 - 11". While more correct given the errors in ranking procedures, it's certainly less fun to say that I'm 90% sure that my department is somewhere between the 5th and the 14th best program in the country using one set of weights for 61 categories measuring aspects of graduate education.