Indeed, research by the economists Eric Hanushek and Steven Rifkin -- both advocates of school reform -- indicates that neither teachers' own test scores when they were students nor their educational credentials explain much of the variation in their students' outcomes. Why judge teachers narrowly on a set of outcomes that are not even predictive of their own success?
What I don't get is the last sentence. Imagine we were talking about doctors and the question was whether to evaluate them based on patient outcomes. Suppose that the success of doctors was unrelated to their own health. Would you then say, "Why judge doctors narrowly on the health of their patients, when doctors' own health is not predictive of their own success?"
Are student test scores a noisy indicator of teacher quality? Absolutely.
Should the evaluation of teachers be undertaken by remote technocrats? Absolutely not.
Would getting rid of the worst teachers improve student outcomes? Possibly. I believe that is Hanushek's big idea. Again, I am skeptical that any intervention has provable, long-term consequences that can be widely replicated.
I think that the best way to get rid of the worst teachers(short of full-on vouchers with no government-run schools) would be to de-consolidate school districts, which would shift the balance of power toward parents and away from school administrators and teachers' unions.
In my opinion, the large school district is one of the most anti-libertarian institutions in the United States today.