Facebook’s ‘Trending’ Algorithm
http://www.wsj.com/articles/facebooks-trending-feature-exhibits-flaws-under-new-algorithm-1473176652
In class, we have been discussing the ways that Google and other browers choose the order and importance of webpages. One of the primary factors in their rankings is the quality of the links going into and out of the pages. Browsers are not the only sites who need to use such a ranking algorithm. Social media sites, such as Facebook, also need to consider how to order posts. This algorithm is particularly important for Facebook’s “trending” feature. This feature is the primary news source for many of Facebook’s users.
Initially, Facebook did not have an algorithm. Instead, it had people curating news stories to highlight in the “trending” section. However, Facebook came under fire this past spring for being biased in its “trending” section; many Republicans claimed Facebook’s featured articles were generally liberal. Although Facebook claimed otherwise, in August, Facebook switched to using an algorithm for its features news. This algorithm did eliminate complaints of bias. However, once adopting the algorithm, Facebook faced even larger issues. Although the algorithm was unbiased, it could not recognize when there were errors in the articles it featured, nor could it discern between real articles and satirical articles. Additionally, because it featured “trending” articles, many of the articles were gossip/celebrity stories, as opposed to actual news. Assistant professor of communication at California State Polytechnic University, Pomona, Jason Turcotte addressed this issue in the article, saying “Algorithms don’t necessarily respond to news values or journalistic ethics. Algorithms respond to keywords, search terms and trends.”
The issues Facebook has had with “trending” illustrate the problems that can arise with algorithms, particularly when these algorithms are hastily developed. When learning about PageRank and other ranking algorithms, it is important to think critically about exactly what this algorithm is meant to deliver. It is also important to acknowledge the role that people’s decision-making can play in improving the results from this algorithm.