So long as it focuses on the creme, then it'll disrupt that 'market' but it cannot disrupt the 'hiring industry' because at that point, it would become diluted and another 'disruptor' would cater to the 'cream'.
Of course the return rate won't be the same with 1 million people. But if you are grameen bank (http://www.grameen-info.org/), you institute mechanisms so keep the return rates high. That doesn't mean that there are no defaulters at grameen bank. It just means they were able to work creatively on a problem and actually meet their metrics (maybe sligtly relaxed) at very high scales.
It's non-trivial, but not impossible.
I think a measure of low quality is: do users/participants actually stop using it because of bad quality? It's a point of debate, but for me HN's quality is as good (probably better) that it was several years ago. Just my opinion.
Seen another way, let's say the top talent is 10,000 individuals. There are way more positions and candidates than that out there. So if they became the Reddit of job boards, they'd have tens (dom)/hundreds (int'l) of millions of applicants in search of tens (dom)/hundreds (int'l) of millions of jobs. Neither the applicants nor jobs are going to become "great" just because they are on the Angellist board.
And the whole site is based on people actually spending tens of minutes crafting reviews. Who'd have thought it'd work out as well as it does?