There's a big missing point in this argument: it says "better" or "worse", "it works" or "it doesn't work", but does not specify how this is tested.

If we test students with standardized tests of their knowledge of facts and simple routines, I'm 100% convinced that direct instruction works better. I'd like to see if it's better also on aspects like student welfare, ability to reason and solve complex problems, creativity and innovation.

It _is_ possible that direct instruction also works better in these metrics, I just think this should be made explicit.