It's somewhat hard to argue with a generalization like this. It's like saying: you give loans to 100 people meeting some very specific criteria, and the return rate is great. When you scale it to a million, the return rate isn't the same any more.
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.
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.