That is wrong on two counts, Adina. Firstly, assessing what is actually played is enormously problematic, and with a Power-law distribution most methods of sampling will significantly overweight the 'top 10' musicians, compared to the larger number with a few plays at the bottom end of the curve.
Secondly those few who are achieving sales of 500,000 will be happier with a per-download payment of a very small amount, which helps squeeze out the smaller creators where the payment does not amount to much. They are more likely to do well from a voluntary system, where the happy customers will willingly pay.