I jokingly mentioned that a recent free software licensing flamewar was approaching the most comments over upvotes, but when I crunched the numbers, it was nowhere near that. Another user asked me for some more details.
This Google sheet shows the top 100 submissions, sorted by comment/score ratio. In HN&&LO I mark submissions with a ratio above 2.5 as “controversial”.
Some caveats: the score is net. I don’t track upvotes and flags separately, which means that a submission with a low net score can have similar amounts of upvotes and downvotes (itself a measure of controversy).
The numbers are from when I gathered the data, and may not be 100% up to date.
I’ve sorted by the absolute ratio, to account for negative net scores. This might skew the results a bit (shouldn’t a submission with -2 net and 20 comments be rated more controversial than one with 2 net and the same number of comments?) but I honestly don’t know how to account for that.
A cursory glance at the dates seems to show that the most controversial submissions were earlier in the site’s history.