The graph shows that as HN traffic has grown, traffic swings have become even larger. This means that the majority of HN growth has been strongly in one-time zo

The graph shows that as HN traffic has grown, traffic swings have become even larger. This means that the majority of HN growth has been strongly in one-time zone. Another observation is that traffic swings have very sharp peaks which means HN ranking algorithms need to be extraordinarily robust to not depend on the availability of huge traffic. Unfortunately, that doesn't seem to be the case as you can see stories that get ranked on top are pretty much random. I think we would be shocked to see qualities of stories that got ranked on the bottom of these daily traffic cycles. I'd say HN is a very nice challenging research problem for collaborative filtering with highly fluctuating traffic.