YouTube’s Algorithm Creates Echo Chambers
https://www.theguardian.com/technology/2018/feb/02/how-youtubes-algorithm-distorts-truth
As the largest online video database, YouTube sifts through the never-ending stream of videos uploaded to their site every hour and creates a subset of videos it presents to audiences worldwide. YouTube then broadcasts these select few videos across its homepage, trending tab, and related videos tab. With this unending stream of videos constantly berating their servers, YouTube prides themselves on the ever elusive “algorithm” which they claim separates out the quality from the quantity. YouTube accomplishes this by showing any given video to a few accounts, and if these videos get watched the algorithm shows these videos to more and more users. Eventually, videos that are catchy, interesting, or otherwise capture the eye of the viewer rise above the rabble of the rest. It doesn’t end there though! YouTube also takes context into account. Say a video is very popular among certain crowds, such as a fashion video, this algorithm also notes that popular fashion videos get more clicks when they are presented after users watch other related fashion videos. Here it becomes clear that YouTube’s algorithm is doing far more than simply lifting popular videos to the top. Rather, the algorithm creates Networks connecting videos that users watch in conjunction. Educational videos recommend more educational videos, comedic more comedic, and news videos promote news videos.
An unintended and unfortunate side effect is the rise of bubble communities creating echo chambers within YouTubes website, as anyone who delves deep into the rabbit-hole of videos can tell. The article talks about the negative political effects of conspiracy theory echo chambers, but more broadly the algorithm seems to feed viewers videos they want to see rather than videos they need to see. This relates directly to class as if we think of the connections between videos like the connections between people, there tends to be a large clustering coefficient between similar videos while there are weak to sparse connections to videos outside the echo chamber. As a result, anyone can open the website now and click a video on the trending tab talking about a recent political event, then click just one recommended video that looks vaguely conspiratorial. Bam! Just like that now every video suggestion and recommendation they receive will be similar conspiracy driven content as they are trapped within the web of interconnected conspiracy videos.