> it's totally possible to (by, say, 2036) train 100% of teachers to perform at a 90th percentile as compared to teachers from 2026. That's how improvement works, which is what people are describing here.

I doubt you can pull this off unless you’re willing (and able) to fire at least 25% of teachers who appear not willing (and under strong unions cannot be required) to outperform the current 90th percentile teacher.

There are great teachers; there are also entirely lazy/entitled teachers who will never willingly be at the performance of the current top 10%.