Sterling Ambivalence (sterlingnorth) wrote,
Sterling Ambivalence

Auditing the Gender Genie...

Over at, they purport to be able to determine the gender of the writer of any passage submitted to it by analysing the choice of words. I decided to test that...

I took passages from two guys and four girls. (Well, I can only confirm the gender of one of the writers, since she was the only one I met in person. For the sake of the test I assume no one is pulling a "Mary Rosh" and is really the gender that they claim -- in mind, if nowhere else. I am going to try to avoid any gender identity issues, for this simple test of a test is not prepared to even try to take transsexuality into account. And damn, I did not mean to have such a long disclaimer...).

Anyway, I took passages from the journals of four women and three men from my friends list. I scanned through the four most recent pages of each journal and took the longest passages from each and fed it into the Gender Genie.

My findings are thus, so far... There is a bit of a maleness bias... when it errs, it more likely errs by reporting that a male wrote a passage when a female in fact did. There's also a less powerful bias in thinking that personal entries (entries about oneself) are written by women... the times it mistakenly classified a male writer as female were all on personal entries.

My experiment wasn't the best and the sample size is microscopic, but their statistics seems to confirm my findings. (See under "Blog Posts")
  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.