There are factors other than teacher's merit that affect outcome. You're saying that a teacher who gets two learning-disabled students (who need to be paired due to sibling issues) is less deserving than their peer who didn't have the disruption of two learning-disabled students?
The moment you tie income to results, you incentivize teachers leaving behind students who won't perform.
It's amusing (and sad) that almost everyone involved with a school can quickly point out the best and worst teachers there, but there's no way to "programmatically" encode it so that the bureaucracy can do something about it.
Because it's not that hard to measure teacher quality. But once a measure becomes a target, it ceases to be a good measure. It gets gamed and manipulated. When it was just a measure, no one bothered to game it, so it wasn't that hard to tell who was good and who wasn't.
Imagine a system where of 500 teachers, 25 that have a track record of doing much worse than average are let go. You then roll the dice to get new teachers.
Could you pay more to attract better teachers in that situation and get better overall value for the kids being taught?
So how do you identify poor teachers? I would guess that if you let each member of the faculty vote for the 25 top teachers and then have parents do assessment of their children's teachers and then average the results over 3 years, you could come up with something that has virtual no false positives as to who the worst teachers were.
Evaluating teacher quality is difficult and impossible to do perfectly. Even a simple effort to evaluate quality would be vastly superior to the current state where quality isn't even a consideration.
I agree with the ensemble of evaluations approach - student tests at the start and end of year show student attainment, student and parent feedback, peer feedback, and administrator feedback. Come up with a weighted average and experiment with it. Retain average and better teachers, reward rockstars, train underperformers, and let the bottom ranks and those who don't improve with training go.
The idea is to randomly assign the students to teachers. Sometimes teachers will get students who will never reach grade level, and sometimes students who will effortlessly achieve. By being random, it evens the opportunity out.
That sounds horrible. Right now, schools try to give each teacher an even number of high-performing, average, low-performing, and “difficult” kids. But the numbers are so small it’s hard.
An average sounds like a shitty, shitty system. Imagine finding out your kid is in a classroom with every low-performing and difficult kid in the grade just because it was random.
The concept sounds good in theory, but I think it's going to be nigh unworkable in practice. The NCLB/high-stakes testing era exposed many problems with tying educator pay to student outcomes -- chief among them that student outcomes didn't improve.
The moment you tie income to results, you incentivize teachers leaving behind students who won't perform.