Google Search Engine, PageRank and Racism: Problems That Arise With Google’s Ranking Algorithm
https://time.com/5209144/google-search-engine-algorithm-bias-racism/
This article is about how Google’s search results are organized in a way that underscores gender and racial stereotypes, calling the algorithm itself “bias against black girls.” It is directly related to some of the problems we discussed in class about search engine algorithms as well as addresses the page rank algorithm that we learned in chapter 14. The author, Safiya Noble, a professor of Communication at the University of Southern California, talks about how when she searched up the term “black girls” on Google, the first pages to appear were primarily pornographic. Even though there was no mention of pornography in the search box, the results that appeared still heavily included details and web pages that led to pornographic pages. Words used to describe black women in these search results were “sugary,” “hairy,” ”booty,” ”porn star,” ”hot,” and “hard-core” among others. The author attributes this ranking of search results to the prevalence of over-sexualized content of black women on the internet and Google’s advertising placement system. Before commenting on the validity of the author’s argument however, I want to first go over how Google’s ranking algorithm works as discussed in the reading.
In chapter 14 we discussed the problem of ranking when searching the web. Google determines how to rank their pages based on the information in the web itself, not by contextual clues about the information. One of the main problems addressed in chapter 14 is that although search engine algorithms can search through millions of pages based on what the user searched, the user realistically only wants to see a few of these. Because of this, Google or other search engine companies have to organize the pages shown to the user in an easily digestible way and ranked according to the most relevant information. However, there is a common misconception held by the public that this ranking is somehow done in a subjective way, where Google manually chooses which results show up. The truth is that the pages are ranked based on how frequently a website or link is linked on other pages. This method of ranking by links is called voting by in-links. As stated in the textbook, it is the hope that because a page is linked many times in other pages there is a collective endorsement of the page, which could potentially weed out pages that have mal intent. If more pages have more links on other sites than this result is considered to have more “authority” than pages that have less.
The article mentions that Google’s search results underscore hegemonic stereotypes that exist in the digital sphere today, specifically focusing on women’s lack of status in society and the over-sexualization of black women. This relates to the way that Google organizes search results because if most digital sites are oversexualizing black women, then these are the pages that are going to have the most authority in Google’s ranking system. In this way, it is evident that Google themselves does not have a bias against black women, but the general information on the internet about black women tends to be hyper-sexual. Additionally, there is one key problem that is addressed in the article and chapter 14: additional advertisement spots. Chapter 14 talks about how search engines can provide additional advertisement slots on the main page of search results to companies that pay the most. These ads are then strategically placed using a matching algorithm. The article mentions that Google will prioritize links that provide them with an opportunity to make the most profit. And, since the companies that have the most money are usually large multinational corporations, often the advertisements that are purchased further underscore existing stereotypes that existed in society in traditional media.
Since this article was published Google has made a concerted effort to censor pages that contain explicit content, yet they still cannot have complete control over what people search.