The scales used on global measures of education are scored in a way that makes a few low scores a lot more influential than a few high scores (since the average score on the test is around 70%, it takes more than 3 100% scores to balance out one 0%). So countries which are more homogenous (so that their upper end's scores don't get capped at 100%) and don't have as big of a low class do well, such as Japan, Korea, and Finland.

If the tests were structured so that the average score was 10 out of 100, one genius who scored 100 would be worth 9 people who scored zero.

I think this is a more realistic way to score education; would you rather have the median of the population go from grade level 7 algebra, to grade level 7.1? or would you rather have 20% more utter geniuses in your society? Both have value, obviously, but breakthroughs are made at the edges, and only need to be made once.

All this is to say that we should focus on where the returns are greatest.