The NY Times was able to show that if you look at the "value add" of teachers who teach two different classes the correlation between their results for the two

The NY Times was able to show that if you look at the "value add" of teachers who teach two different classes the correlation between their results for the two classes is very weak, suggesting that the value add measure is nearly worthless. (You're incentivizing luck.)