Companies have invested significant resources into developing advertisement algorithms that analyze users’ data and display relevant media in their feed. People are more and more shocked that the products they searched for now appear on the tabs of their social media and email accounts. Other than proprietary reservations about one’s data, this 3-way relationships is seemingly symbiotic; Ad companies’ services are being shared with costumers most inclined to buy-in, users see more of the product they like, and the company generating these relationships reaps a hefty profit.

But how do these algorithms work, and what are the effects of seeing only the information companies like Facebook and Google expect you to like? Companies use machine learning programs that try and identify common phraseology and links as the parameters for classifying similar content. The words you tend to click on will be the ones you tend to see. The more a user clicks on media that have a similar progression of words and ideas, the sharper the computer’s vocabulary becomes, and it will anticipate which new articles and websites the user would herself visit. And while this makes social media browsing all the more pleasant, there is a severe impact on the variety of perspectives a user sees.

News articles are some of the most common media that is filtered by these algorithms. Politicians cater their campaigns to the political perspective assumed by a user’s data. This means that voters are only seeing a very small fraction of larger political conversations and are forming opinions without all the facts. Repeated rhetoric is literally what baits the algorithm, so users will only be suggested articles that have nearly identical viewpoints. This motivates journalists to look for fiery phraseology that will increase their shareability, often at the expense of nuance. As a student, I would like to speak to how this has affected intellectual conversations on university campuses, where students are generally exposed to limited perspectives with much less appreciation for complexity.

In conclusion, I am not arguing for the moral correctness or incorrectness of this methodology for displaying media, but rather I want to raise awareness in the data mining community about the larger impacts of data analysis in the policy worlds of economics, sociology, and politics generally. Data will affect how today’s youth form decisions, and serious threats to diverse exposure are posed when feeds are too severely exploited.