Essentially, it's a combination of time and votes. The older something is, the less it is promoted; the more upvotes a submission has, the more it is promoted.

Essentially, it's a combination of time and votes. The older something is, the less it is promoted; the more upvotes a submission has, the more it is promoted. The original calculations (I believe) can be found in the Arc source code. Having said that, pg (et al) have tweaked the calculation over the years. Some changes are reasonably well-known or clear (eg, longer comments fall slower than short comments; Ask HN submissions fall faster than links); others are, I'm sure, hidden away to prevent gaming the system. Could it be improved? Constantly, I'm sure, in subtle ways. The better question is 'Is it being improved?', to which I think the answer is yes. I'd be interested in hearing more about what you mean by 'page views'. My immediate intepretation is that page views aren't important - a page (link or discussion) that has 10x as many views but fewer upvotes is being measured on quality (upvotes) not quantity (which, if nothing else, can be gamed by a linkbait heading). But I may have misunderstood your suggestion?