While I disagree with Connor's suggestion to let teachers decide which students get to be in their classroom, I think he is correct in pointing out that the teachers that are most preferred by the best students are likely to be "the best", and therefore deserve to be rated accordingly.

Fascinating blog. I think the problem with pre-college education is 70-80% related to the students' home environment, and only 20-30% related to bad teaching, which is why I am skeptical of performance-based payment schemes as an adequate solution. The challenge for policymakers is how to make success in school something appealing for students, rather than a burden that they must carry in order to succeed later in life. The current system seems to assume that kids with rich and poor backgrounds respond equally well to improvements in teaching, which they certainly do not. Kids with privileged backgrounds (i.e. high IQs, peaceful neighborhoods, caring parents, etc.) are likely to thrive in a system that rewards the teachers that improve their average standardized test scores, since - as Posner mentions - they are "likely to improve more", and their peers are more likely to have a positive effect on them. Kids with poor backgrounds need some extra motivation to make the best of the school experience, since their environment often offers them alternative (easier) ways to improve their "social status". In the best of cases, they are lured into the arts or sports, in the worst they resort to crime. Hence, perhaps we should consider giving "merit pay" to students rather than teachers? It may even improve their home environment, and make their parents more interested in their child's education.