Earlier quoted context omitted.
It's actually not obvious that a testing-based teacher measurement regime would be better than the seniority system. The seniority system does lock in mediocre teachers, but it has minimal knock-on effects to the actual curriculum. Test-based measurement has already made some curricula toxic to students, and those test results have minimal impact to the teachers themselves. How much worse would it be if the teach-to-…
I'm guessing less worse than the incentive to do the minimum amount possible knowing you're going to get a pay raise regardless. We can change the test if it's not measuring the right thing (we should do that anyway regardless of it's use for teacher measurement). Test scores are what parents, universities, etc. measure the performance of children by, we shouldn't have a double standard of having one metric for child…
Plus, universities do take some other things into account. I got significant credit on my university applications for an open-source project I worked on in high school, and a website I maintained (in the late 90s websites weren't exactly rare, but it wasn't yet to the "everyone has a blog" level of pervasiveness). Will a teacher who successfully inspires kids to do that sort of extracurricular stuff get credit in these evaluation schemes?
Alternatively, if you consider preparing kids for college to be the sole purpose of K-12 education, why not just measure it directly? For 2nd-grade teachers it would be difficult to measure due to the large time lag, but it'd be possible for high-school teachers. Instead of the state inventing its own tests, just use the % of students who go on to college as the metric, and then do some statistical analysis over a 5-year window to estimate the effect a particular teacher had on that rate. There are a lot of variables and each student is influenced by many teachers, but over a window you can make an estimate of whether a particular teacher had any influence (positive or negative) on college-admission rates, compared to a hypothetical average teacher of the same subject and of a hypothetically equivalent set of students (according to whatever demographic/etc. variables). A teacher version of http://en.wikipedia.org/wiki/Value_over_replacement_player, basically.