2 ms·

I just looked at a HN comments dataset, for May 2018. Empirically the number of commenters, c, needed to "start x%" of conversation is sqrt(c) %. So 144 commen

I just looked at a HN comments dataset, for May 2018. Empirically the number of commenters, c, needed to "start x%" of conversation is sqrt(c) %. So 144 commenters gets you 12% of "conversations started". Where I define that as there being 144 commenters whose comments are the parents of >=12% of all comments. At 2500 commenters, 50% of all conversations are accounted for, etc. Note, this is just an empirical formula for 0 to 3k commenters on this dataset, ie., its just very close to sqrt(c) for this range. The general formula will be different, but approximated by sqrt(c) within a small range.

h2odragon 5y ago
Interesting statistic, but does it mean that much? A board where the same 144 assholes agree or argue with each other all the time isn't likely to be that much fun for anyone else. 2500, not much more so. The community is the thousands of people that make up the long tail and usually say very little, individually. The sufferers of verbal diarrhea who stand tall in your statistics are just something you get in any community, not necessarily reflecting a larger opinion or leading it, just decorating.