I feel that this experiment can prove nothing about the quality. It does show that votes from initial users have roughly same distribution as votes from later u

I feel that this experiment can prove nothing about the quality. It does show that votes from initial users have roughly same distribution as votes from later users, but only because the initial exposure to the front page highly correlates to subsequent votes. So it rather proves that both group of users have the same usage pattern: mainly watching the (non-classic) front page. It would have been more interesting to see the front page that only consists of submissions from initial users.