Comment by lostmyacctoops
3 years ago
I'd be very curious to know if these algorithms can link very different types of text. I'm not surprised that my style is "derivable" on HN, but what if you included my slash-fic pieces, my research papers, etc, would it still "catch" me?
Also, talk about a chilling effect. I was already vaguely aware of this, and now I'm overthinking every word I'm thinking/typing.
I'm gathering that they just took a bag-of-words approach to this; basically comparing word frequencies. Writing across content types (fiction vs technical writing for example) will probably show different word frequencies, especially technical jargon, and so on. More sophisticated approaches are possible.
And yes, potentially very chilling. If you want to post truly anonymously, you might want to run your words through some kind of filter first.