Yes, I meant the subject tests, when talking about the high score range. They are still scored up to 990. However, because of the way the mean is set, I think it may not be possible to score 990 in some subjects. The other point is that the GRE subject tests in physics and math tend to cover only the more elementary material. CB claims that it is the first three years, but it looks to me like the first two years.
I think the general GRE tests are useless as a measure of undergraduate learning. A strong high schooler could score quite well on them.
In my view, the right way to assess an American undergraduate education is through the transcript (and if it were possible, the actual syllabus for each of the courses). Back when I applied to MIT for grad school, they wanted a list of all of the textbooks we used for each science/math course. This makes good sense to me, although it is imperfect in assessing the course level.
At one point, I had the students in a statistical mechanics class calculate how many different sets of courses could be taken at my university, to exactly fit the graduation requirements with regard to number of credits (not guaranteeing that the general education requirements would be met, or that the requirements for a major would be satisfied, or that the selections would make any sense–just the raw number). It came out somewhere around 15 trillion. While that is obviously a wild overestimate of the possibilities once you include the constraints on course selection, it is nevertheless the case that most large universities (and the very top schools) have many, many routes to graduation that are pretty easy and many, many routes that are pretty hard. The courses that a student selects (when given any choice) give a good indication of the student’s level of thinking, so the transcript useful in that way also.
GPA is not a pointless indicator when coupled with the transcript, but separated from it, it doesn’t tell you much.