Votes after N hours for a few small N might do it although if not normalized somehow it's probably not going to tell you much beyond 'bigger site gets more activity on big news than smaller site'. Maybe divide by average daily votes at the time?
Publish the timestamps of all votes for the top 10 most upvoted stories. Then the community can create scatterplots showing the acceleration of each story's score:
(def allstories ()
"All visible loaded stories"
(keep cansee (vals items*)))
(def mostvoted (n (o stories (allstories)))
"N most upvoted stories"
(bestn n (compare > len:!votes) stories))
(def votetimes (s)
"The timestamp of each vote, in ascending order"
(sort < (map car s!votes)))
; save vote timestamps for top 10 most upvoted stories
; each line contains the story id followed by a list of timestamps
(w/outfile o "storyvotes.txt"
(w/stdout o
(each s (mostvoted 10)
(apply prs s!id (votetimes s))
(prn))))
; paste storyvotes.txt to https://gist.github.com/ and post the url here
Note that this prints the timestamp of all votes, whereas each story's score is vote count minus sockpuppet votes.
If you don't want to reveal the timestamps of every vote, you could randomly drop K timestamps for each story, where K is the vote count minus the score. (E.g. https://news.ycombinator.com/item?id=3078128 has 4338 points, and you'll only reveal 4338 timestamps.) Since there are thousands of votes, this won't skew the scatterplot much.
