I note that the /old experiment (a Hacker News whose voting data set was limited entirely to accounts > 1 year old) demonstrated no obvious improvement in quality.
It seems to be entirely a question of the public choice of what is acceptable content to submit.