The best predictor for success in physics is still how much math knowledge a student has.

Wholeheartedly agree with that one.

But I can't decide if it's time to evolve the teaching style a bit. By and large, the current format still consists of timed testing at a desk with only a pencil and paper. I've gotten serious test anxiety from this method, and it corresponds to few (if any) work scenarios after receiving a diploma. On the other hand, knowing that I have to do physics with a gun to my head sure does motivate me to hit the books.

I have been seeing more open book and take-home tests as I've progressed through my education though. But the level of the curriculum has to be sufficiently advanced such that the answers aren't google-able.

Had one prof this semester give an exam that was impossible to finish more than 60% of even if you knew the curriculum backwards and forwards. Yeah, the grade distribution will be re-scaled, but I've already been struggling with feelings of inadequacy. I'm actually under the impression there is a national culture of inducing existential crises within grad students.