Search Bias

Learn about how algorithms contain bias and affect search results even in trusted information sources.

Today you feel ready to search. You are confident that you know what it means when a source is scholarly. That is, it's found in a peer reviewed journal or other publication. It has a list of references and it offers a good argument. You've been to a library search session, a time or two, and know that the best way to start searching broadly is to use a database. These assumptions are all correct, but they're not the end of the story. Unfortunately, databases are actively hiding important materials from you using something called algorithmic bias. But there are tools and strategies you can use to make sure you combat this unintentional but pervasive problem. What is algorithmic bias? In his book Masked by Trust: Bias in Library Discovery, Matthew Reidsma explains that algorithms are examples of pattern recognition that show up when many, many searches are run in a database or search engine. Algorithms are deeply woven into most aspects of our lives, from Netflix suggesting what you'd like to watch next, to Facebook promoting posts that are more prone to listening and emotional response. Yet these algorithms, promoted as curating perfectly customized recommendations, are thoroughly human. Software engineers constantly update their programs to improve performance. Sometimes experiments are run without users knowing. Remember when Facebook promoted upsetting material to one group of users and uplifting material to others to gauge which type of content promoted the most clicks. And if it's really totally customized, can you and I even say we use the same algorithm while Googling best restaurant in Lawrence. But you might argue databases claim that they're unbiased, influenced only by the search terms and other parameters that you select. And it is true that library databases are free of certain kinds of external influences like advertising, the harvesting of all your browsing data to curate resources, and the lure of designing material to show up first on a search engine, otherwise known as SEO. But the tools you use to access materials on campus is selected, arranged, catalogued, and made accessible by the same folks who designed algorithms. Human beings, complete with all their internal biases. The use of keywords as neutral is highly contested by librarians. Consider the constant review and revision of offensive keywords that is undertaken regularly by the Library of Congress. In addition, as Reidsma notes, they assumed that broad topics can be easily identified by a few keywords and reference material is always reliable knowledge source. Aside from the problems of assuming the material gathered is neutral. What about the nature of lived experience and the information that grants? What about traditional knowledges? What about oral histories, zines, blogs. All kinds of places where knowledge making occurs outside the academic framework. Of course, knowing these pitfalls doesn't mean that you can or should discard these resources. But it's good to be aware that a database search does not provide all the information that there is. Ask yourself when searching, who is the audience? Who benefits from this information being selected to promote? Who is silenced? Now that you've answered those questions, consider those authors and keywords. Try and make sure you're looking for underrepresented authors and perspectives. Use multiple databases, really large databases like Academic Search Premier, or ones that contain a vast amount of information. But other databases are available that offer a small but extremely thorough collection. In particular, checkout ones that offer alternative perspectives on a topic. Read more about algorithmic bias, along with Masked by Trust, checkout Algorithms of Oppression by Safiya Noble, Race After Technology by Ruha Benjamin, and The Intersectional Internet edited by Safiya Noble and Brendesha Tynes. Most importantly, stay alert and stay critical when searching. Never trust that the perfect resource exist. Use the same skills you've cultivated in evaluating resources to consider the biases and strengths of how these resources get it to you. Talk about it with your professor and other students. Know that a librarian is always ready and excited to talk more with you about it.