I think the key take-away is they're using evidence to make decisions. There's no end to prescriptions in education: small class sizes, start early, selection, non-selection, child-led, and so on.

From my own experience, it makes sense to make less of the tests. I liked tests, and I did math contests, but a lot of people don't like it. And when you look at work, you rarely have an occasion where it's just about knowing something and doing a calculation, which is what they do in high school maths. The times I've done math at work it's been something like this:

- We have some option model that needs to be implemented

- The model turns out to require a function that we don't have in the standard lib (Bessel function)

- There's a book with an implementation in C

- There's a wikipedia page that we'll scan to make sure there's no gremlins

- We then start to write the implementation in our own lib

- We do some tests to make sure the number that comes out of the Bessel function is correct

- We do some tests to see that the option model comes out with the right number

None of that workflow is anything like in school, where you'd take some function, apply some trig relations, apply the list of integration rules, and plug in the starting value. Without a book.