I would like to see a scoring algorithm that dampens (if not totally eliminates) time-of-day effects. Essentially, clock time shouldn't be a factor at all: just some other synthetic 'tick'. That tick is probably 'sitewide votes' (but might be 'legitimate votes' or 'net upvotes' or 'homepage views').
Secondarily, I would like to see a decay function that gives an older story with a current surge as much credit as a new story with similar recent surging.
(For example, imagine story A submitted at hour 0 which receive no votes for 6 hours. At hour 6, story B is submitted. Between hour 6 and hour 7, story A gets 12 upvotes, and story B only 8. Over the current-comparable period, A is hotter... but I'm pretty sure the HN algorithm and most similar sites penalize A for the earlier-submission time. And yet due to things like headline corrections, the eventual arrival of a story's natural audience, etc. the A story may in fact now be 'better', in those attention-deserving qualities that the ranking/decay function is trying to detect.)
As long as the quantity of good submissions varies consistently depending on time of day, there will be times when more good submissions are competing for a fixed amount of frontpage time. Thus it's unlikely you could eliminate time-of-day effects.
Secondarily, I would like to see a decay function that gives an older story with a current surge as much credit as a new story with similar recent surging.
(For example, imagine story A submitted at hour 0 which receive no votes for 6 hours. At hour 6, story B is submitted. Between hour 6 and hour 7, story A gets 12 upvotes, and story B only 8. Over the current-comparable period, A is hotter... but I'm pretty sure the HN algorithm and most similar sites penalize A for the earlier-submission time. And yet due to things like headline corrections, the eventual arrival of a story's natural audience, etc. the A story may in fact now be 'better', in those attention-deserving qualities that the ranking/decay function is trying to detect.)