Facebook and PageRank
Sources:
Constine, Josh. “How Facebook News Feed Works.” Tech Crunch, Tech Crunch, 6 Sep. 2016, https://techcrunch.com/2016/09/06/ultimate-guide-to-the-news-feed/.
Kincaid, Jason. “EdgeRank: The Secret Sauce That Makes Facebook’s News Feed Tick.” TechCrunch, TechCrunch, 22 Apr. 2010, https://techcrunch.com/2010/04/22/facebook-edgerank/.
McGee, Matt. “EdgeRank Is Dead: Facebook’s News Feed Algorithm Now Has Close To 100K Weight Factors.”
Marketing Land, Marketing Land, 16 Aug. 2013, https://marketingland.com/edgerank-is-dead-facebooks-news-feed-algorithm-now-has-close-to-100k-weight-factors-55908.
As with any social media platform, Facebook’s primary goal is to keep users engaged and interested. As Constine explains in his article, “[Facebook] wants to choose the best content out of several thousand potential stories that could appear in your News Feed each day, and put those in the first few dozen slots that you’ll actually browse through.” As such, these stories (posts, photos, articles, events, etc.) must be ranked in order of importance to the user to ensure the user gets to see the stories most interesting to them. For instance, the user is more likely to be engaged in the posts/events of family and close friends that they interact with on the daily. They are more inclined to read, like, and comment on these stories as opposed to more boring stories such as a distant acquaintance liking a news article.
Facebook does this by assigning each story a relevancy score and listing content in order of relevancy on a user’s News Feed: an implementation of the PageRank algorithm we discussed in class. As Constine mentions, the algorithm must take into account countless factors, some of the most important being: who posted, how other people engaged with the post, what type of post it is, and when it was posted (time decay). If each story was treated as an authority, these factors could be considered hubs contributing to the score of the post. The hubs (different types of user engagement) act like a feedback loop and in turn update the authority of posts.
Though, this barely covers the complexity of Facebook’s PageRank algorithm, originally called “EdgeRank.” It is constantly evolving. McGee speaks to how EdgeRank is now very outdated, as “the company began employing a more complex ranking algorithm based on machine learning.” It currently takes into 100k weights which are also constantly changing. The original EdgeRank model is in decline. It only relies on a few static factors that are not personalized for every user. The extended machine learning rank algorithm “only ever gets more complicated,” taking into account the user’s likes, clicks, etc.
The analysis of Facebook’s PageRank algorithm also lends insight into the Ad Market. In order to generate revenue, Facebook also includes ads in one’s News Feed. As we discussed in class, websites and search engines such as Google and Bing get money for an ad based on the number of times the ad is clicked. McGee speaks to this, as “Google and Bing have a lot of new signals, like personalization” to ensure these ads are clicked on. Similarly, Facebook targets specific ads at users that the user is most likely to click on. Again, Facebook uses a (different) PageRank algorithm that works in the same manner: taking information about the user such as likes, dislikes, and interests to determine which pages and business ads to present on the News Feed. With the help of this algorithm, Facebook can make billions of dollars every year solely through advertisements.
Ultimately, much of Facebook’s success, including 1. user satisfaction/retention rate and 2. financial revenue from the Ad Market, can be attributed to its PageRank algorithms.