A few years ago, I reverse-engineered the HN ranking algorithm [1]. Basic ranking is based on votes vs age, but there are numerous other factors. Penalties are applied based on words in the title or the domain. Posts with too many comments got penalized as controversial. A "voting ring detector" triggers other automatic penalties.
HN ranking has probably become more complex since I looked at it. Manual moderation also impacts ranking.
I suspect the formula is changing if only because the sheer volume of spammy type stuff skyrocketing to the top seems to have subsided.
There was a short run where a lot of nearly raw spam type posts were quickly rising to the top, and sticking a long time despite almost / all the comments being about how terrible the spam is.
Good analysis, but - obviously - outdated and besides, the biggest input and wildcard is 'what are the penalties' and without knowing those the analysis is not super useful even if it is accurate to a rough approximation.
Keeping the spammers out is already quite a bit of work so I'm perfectly OK with the secret sauce staying secret.
