Computer Bias

SS is working on her masters degree in information studies (aka library science). When we have a few minutes to chat, she'll often mention some of what she is reading for her classes and what they are discussing. A couple of times, I hadn't really thought about something that was being discussed. Most of the time, I remember similar discussions from my grad school days and we can even talk about how it fits into my practical experience in the library. It's satisfying to be reminded of how I have managed to integrate so much of what I learned into practical application (and to be maybe a bit of a mentor on a subject). :)

Today she brought up an article the class was reading and preparing to discuss on computer bias. The article discussed and gave examples of three different types of bias--reservation software preferring to book a single airline for an entire trip over cheaper prices (technical bias), systems for placing medical students in residencies giving preference to a single partner between married medical students (emergent bias), and credit ratings being affected by home address (pre-existing bias). While she agreed it was somewhat interesting, she wasn't sure how it related to information studies.

While I don't know exactly which direction her professor will take the idea, I could explain how bias also shows up in research databases. Bias isn't always noticable, because usually it does the job it is supposed to and helps researchers find the results that are wanted. For example, a keyword search in most library catalogs will first return the results where the terms are in the title (or maybe even the subject) of the book. Lower down in the results list will be items where the term shows up in the abstract or the table of contents for the book.

In other cases, a bias may entirely change the results that are returned. In one particular database, all of the search terms entered in one search box are searched as a phrase--without any additional designation as a phrase. For the unaware searcher, this can severely limit results or even return no results--which is deceiving and frustrating.

For a librarian to be able to best help their patrons, an awareness of such biases is important. Finding something in a computer database is not only matter of knowing what is in the database so that one can look in the right place, but also of knowing how the database is searched so that the best information can be found.

Trivia: I believe I remember that in the early days of Google, it would only perform phrase searches. Anybody else remember this or have confirmation that this is true?