The author discusses how educational reformers have encouraged states across the country, as a requirement to be eligible for Race to the Top money, to measure teacher and school quality through value-added measurements. The idea behind value-added measurements is that a computer can take all the information about a class of students and determine what those students would score on the state exams ten months from now if they were taught in an "average" school by an "average" teacher. Ten months later the students take the test, and based on whether their results exceed or fall short of the computer's prediction, that school is given an "A" or an "F" rating, and the teacher is correspondingly rated as effective or ineffective. Unfortunately for schools and teachers, the computer isn't very good at predicting what it is supposed to do. With value-added, teachers rated highly effective one year might be highly ineffective the next despite, according to the teachers, the fact that they didn't do anything wildly different. The author suggests that the state tests are not very good, and threatening a teacher with termination or a school withturnaround unless they measure up on an unreliable scale is never going to improve education.