> The system acts to identify and foster them, but to the detriment of national averages.

This is the usual argument against stratified education systems but is it proven in any way? My experience with stratified systems is that they increase both high quantiles and the average. Maybe (even if not established), maybe they reduce low quantiles like the 0.1th, but very unlikely they damage the average. Quite the opposite.

It's kind of like this idea that communism makes an average citizen richer which sounds logical, but in the end everyone's poorer. Remove the stratification and there's no incentive to do anything.

> is it proven in any way?

The figure in the article claims to show that this is indeed the case -- though as I noted in my other comments, the y axis is actually uninterpretable ("percentiles" can't be negative).