3 ms·
I picked several suspicious words and ran them through searchyc. The four rarest words in the post are: pure-ad, CTRs, disinformation, and dissimilar. Walked
by JesseAldridge 17y ago
I picked several suspicious words and ran them through searchyc. The four rarest words in the post are: pure-ad, CTRs, disinformation, and dissimilar.
Walked through the results for each word, pulling in all the usernames:
intersection of: pure-ad dissimilar
['noodle']
intersection of: CTRs disinformation
['jacquesm']
intersection of: CTRs dissimilar
['ivankirigin', 'patio11', 'strlen']
Of those, strlen's writing style seems to be the closest match. So I'm changing my guess from randomwalker to strlen :)
- patio11 17y agoIt wasn't me. Nice approach, though. Just intersecting word choices has very little to recommend it for industrial scale author identification but for a small-ish community like HN it might work, and of course it is trivial to implement if you already have the data source lying around.