You can imitate someone's writing style with a travesty generator but that doesn't mean everyone has a "writing fingerprint". Very few people have a distinct enough writing style to weed out false positives, and since it's so easy to imitate you'll never really know your precision.
I have a few writing ticks (parens, the '--', certain words like 'certain') but it's much easier to just search on "aristus" to start exposing my shame.
I played with this a few years ago with a project called unmaskr. Heuristics can help precision a lot but does not help with recall. People generally:
* use similar usernames, or a "constellation" of usernames
* have semi-regular posting times
* post in one place at one time
* use similar place names, nicknames for things
* write fluently in one language
* link to a "constellation" of domains
I have a few writing ticks (parens, the '--', certain words like 'certain') but it's much easier to just search on "aristus" to start exposing my shame.
I played with this a few years ago with a project called unmaskr. Heuristics can help precision a lot but does not help with recall. People generally: