If I were to single out one thing which holds much of US education back from excellence, it is the structure of tests.
So much of the testing used is multiple-choice, fill in the bubble style, which prevents students from being tested on developments in the Mathematical areas, and from expounding on ideas otherwise.
I think back to the tests that I took in French and German at O-level (age 16), which required translation in both directions, making up a story in the foreign language, reading a passage in the language and answering questions in English, taking dictation, and having a conversation. The English language test was similar, with the translation replaced by having to precis (summarize) a long passage in 150 words or fewer. This latter was a very valuable skill, and one which many people would benefit by learning.
Our Science and Mathematics tests required solving long problems, writing essay descriptions of experiments, and developing theories.
How could multiple-choice tests possibly be that challenging?
To be fair, or possibly reasonable, the tests were set up so that 35% or so was a pass/C and 70% was an A.
And then, when all is said and done, we have students who have had their parents/software/tutors do most of their work throughout high school (and sometimes college as well), yet cannot pass tests. Those students are then claimed to 'not test well', and are given special accommodations. We had cases of assistants provided by of Office of Accessibility simply reading the answers to students, line by line, during tests. When the latter happens, what is the point of testing or grades at all? When those enabled students finally hit a brick wall, the claim is that “tests don't check for learning”, when they most certainly can and do.