Google Has PageRank? Facebook’s Got EdgeRank!
“[Your Friend’s Name] likes [Your Rival’s Name]’s post on his Wall.”
“[Girlfriend’s Name] and [Random Guy’s Name] are now friends.”
Eye-widening and curious information such as these pop up in the Facebook News Feed Sidebar with fresh and juicy news every few seconds as you surf Facebook. Have you ever noticed in the feeds how a few specific friends pop up in the sidebar more often than others? Perhaps those are your closest friends who you interact the most with on Facebook? Considering that Facebook seeks to increase your clicks and time invested in the site, such personalized news feed system makes sense. But how does it work?
Approximately a year ago, Facebook released bits of tips and insights on how their content rating system, EdgeRank, functions. In class, the concept of Google’s PageRank was explored in which scores are computed based on the incoming and outgoing links the webpage is affiliated with. While such a ranking system can sufficiently isolate relevant search results, in a social environment such as Facebook, mere links are insufficient to successfully evaluate what information will be the most relevant for the users. That is because unlike in Google, Facebook users define their space by creating links with one another through “Friending,” which does not guarantee or appraise the actual proximity in relationship of the users to one another. Therefore, to deliver the information that will captivate the users the most, Facebook developed EdgeRank system.
EdgeRank selects information to place on a user’s news feed through three components: affinity, weight, and time. When a user creates an object (hereon referred to as “input”) such as photos and status updates, Facebook gauges the importance of the input by the interactions of the content that the input generates. How many people “Liked” the content? Were there comments left? How many people viewed the content without any interaction? The statistics on the interaction is counted in this fashion to create an affinity score, which becomes higher the more interaction occurs. In order to prevent the content creator from rigging the score, the interaction is only one-sided such that user A’s actions, such as leaving comments in user B’s Wall, do not at all affect the affinity score of user A’s content. So only user B’s views, posts, and “likes” on a specific input that user A puts up would contribute to the affinity score of that input.
We all prefer visuals over passive texts just as the cliché “a picture is worth a thousand words” goes. Facebook understands this natural propensity as well, and, therefore, gives varying weights to the inputs based on their content types. Although Facebook itself never disclosed the content type priorities, an independent research suggests that the priorities of contents, i.e. higher weight, is given to photo/video, links to other contents, and status updates in that order. Have you ever noticed how links to other contents usually give thumbnails of link destinations? The feature is precisely aimed to provide visuals so to attract more users and, therefore, be of higher priority than the text only status updates. To put this into perspective, the EdgeRank score of a photograph of Big Red Bear recycling water bottles with two comments would outweigh that of a status update on how wonderful Ithaca’s weather was with two comments because, although affinity scores are equal, the weight scores differ. In other words, your sidebar news feed would show the comments made on the photograph, but the ones on status update may well be ignored.
Relevance of information also highly depends on the time that has elapsed since the input was made. For instance, a video of 2008 Homecoming football game uploaded 3 years ago would mean little if it showed up in the news feed today. To prevent that, EdgeRank takes into account the third component: time. Simply put, the EdgeRank score of the content decreases with every second that ticks by, and by the time the input is a day or two old, it most likely would not appear on the sidebar unless a new comment re-raises the EdgeRank score through increased affinity score.
Just as Google’s PageRank gauges content’s importance by providing a score, Facebook’s EdgeRank does the same albeit with different a focus. PageRank focuses on providing the most relevant information from a pool as vast as the Googlebot can crawl on the web while EdgeRank is more interested in the same end result in a closed environment that the user defines through his/her friendship network. Therefore, new components fitting for the different environment had to be incorporated for EdgeRank and those are affinity, weight, and time. The rises of PageRank and EdgeRank in Google and Facebook show that different networks require different systems of measuring relevance and importance. What kinds of input do the users contribute? What is the range of relevant information that the users seek? Combinations of these factors define the needs for different ranking systems, and it may not be far off in the future before social networking optimization services begin to pop up to cheat Facebook as many search engine optimization companies attempt to do so to Google today.