I don't know why you have to postulate "magic". I didn't say that a program that helps each kid progress as quickly as possible wouldn't reduce the variance.

I don't know why you have to postulate "magic". I didn't say that a program that helps each kid progress as quickly as possible wouldn't reduce the variance. It probably would, as the kids on the bottom have more room for improvement. But I would still expect a large amount of variance. Actually, I just noticed that the "after" graph is clustered near 100%, so it's possible that there still is a large variance, but it's just not measured by the tests that were administered. These are probably standardized test scores and don't really address higher end performance. So I guess my points are: 1. I'm doubtful that the "paint by numbers" approach is what the young Picasso needs. 2. Maybe I'm wrong about (1), but we certainly shouldn't subject the advanced students to this method until higher end performance has been measured. 3. I find the attitude of the author, who used the phrase "even things out" as if it was a good thing, concerning. Evening things out should be an explicit non-goal of education.