[educationdata]

When I made that claim, I meant at the national /societal level. Of course I didn't mean there is no single successful education experiment.

On the contrary, if you search through education literature, you will see countless successful programs, but why don't we see their effects at the societal level? Because it almost never scale. Education program is not something you can duplicate easily, it is almost always extremely localized.

In education, it is very difficult for a successful experiment / program to scale vertically (expand to all grade levels) or horizontally (expand to other schools, districts). Also treatment effects are difficult to retain after students left the program, or after the experiment ended.

Interestingly, Taleb just made a twitter this morning: "What works on a small scale almost NEVER expands to large scale." https://twitter.com/nntaleb/status/1035121052886728704