Wednesday, August 11, 2010

Randomization In Politically Sensitive Topics

When I was in high school we had a controversy over whether minorities were and should be over-represented in the pictures of school publications. Some thought the pictures attempted to display a false image of the school to spur donations, and this understandably annoyed them.

It was a tricky question because the school could always claim ignorance. They could argue that any deviations in the pictured sample from the total student population were merely statistical anomalies. Who could logically prove them wrong?

With hindsight, the best solution would have been for the school to draw up a list of all the students in the school and randomly choose which ones to include in the photo. And generally, with the now widespread availability of random number generators, it seems to me that the best solution to representing large communities with small sample sizes is to randomize the selections.

In a related issue, while describing some "athlete" in a post from two days ago I had to choose between the pronouns of him, her, him/her, or one of the many gender neutral pronouns. Choosing just one of the two could subconsciously bias readers into associating a particular gender with a particular activity, perpetuating stereotypes. On the other hand, him/her is unwieldy, and readers likely wouldn't understand the gender neutral pronouns. I was stuck between being a sexist or a sloppy writer.

So I decided to randomize my choice between "him" and "her." I went to, assigned "1" to "him" and "2" to "her," and generated a random number between these two. I got a "2", so the pronoun I used was "her." This was kind of time consuming, but in the future one might imagine word processors offering the randomization of these pronouns as a standard feature. By randomizing, the reader is no longer systematically biased to associate a certain behavior with a certain gender.

One remaining issue is that people might not trust that authors and institutions have actually carried out the randomizations, but instead faked it and went with the option that made them look the best. Thus I can foresee the creation of The Institute Of Randomness, whose role is to impartially randomize words and samples for institutions, authors, and ad agencies. The institute might even offer to randomize the gender, sexual orientation, attractiveness, ethnicity, and etc., of characters in stories, so as to further minimize stereotyping.

Another good part about randomizing is that it eliminates the condescension often involved with direct reversals of stereotypes. Instead, you end up with equality under the laws of probability, which is as it should be.