There is a way for evaluations to be done by the most knowledgeable party and be guided by objective data: 1) Students rate their teachers. 2) Weight each student's input by how well the student improves in testing of overall knowledge relative to peers. When a student has done better, which teacher's are given the credit? A teacher may have motivated a student to do better, improving that student's scores in areas not in that teacher's subject area. Who has had a band director or coach inspire them? "Peers" needs to include students at the same score start value. This prevents advancing teachers who are only adept at improving already good students. A teacher now has motivation to maximize improvement in any student in any subject matter. There is less motivation to "teach to the test". Turning your students into automatons that can fill in the right bubbles gets you nowhere - they will be bored to tears and rate you personae non gratae. Further observe scores over time and match performance back to teachers in the past. Teaching good essay writing in late grade school may not appear on standardized tests, but may show measurable benefit in early college. The biggest known unknown is how the student ratings underlying this can still be abused or distorted. If we cannot solve that, I'm with Posner.