Combating Fake Views on YouTube
https://theoutline.com/post/3804/inside-youtube-s-fake-views-economy?zd=1&zi=46aprclr
Anybody who uses social media for an extended period of time will notice the presence of bots. Spam accounts that post large amounts of content—oftentimes stolen or completely nonsensical—with the aim of generating revenue. However, many are not aware of an entire industry that revolves around inflating the number of “likes” and other metrics to increase the visibility of their customers’ content. This exists on virtually all social media platforms. Instagram “like” bots, fake Reddit accounts with high “karma”, and, as this article discusses, view farms on YouTube.
YouTube often changes its video algorithm to combat unfair practices, but in general the number of views and the viewer retention of each video are positively correlated with how often it is recommended to other users. With this and the sheer amount of content that is posted to YouTube every second, there is a strong incentive for people to cheat the system by purchasing illegitimate views. Social Media Marketing (SMM) panels exist as suppliers of third-party views. Users have the ability to purchase views for relatively low costs (several hundred thousand for less than $100) due to the fact that SMMs can generate views without needing to pay real people to watch the video.
YouTube combats SMMs by trying to determine if the views received by a video are from legitimate (i.e. human) sources. People share videos with their friends on social media. Individuals who are a small distance away from each other in a social network are more likely to interact with one another because they oftentimes have more mutual friends. If a social network is drawn as a graph, as with the Liberal/Conservative graph shown at the beginning of the semester, it is easy to that friends are more likely to share opinions and thus consume the same content. If there is a sudden, large influx of views on a video, YouTube can check whether the views came from accounts that frequently communicate from each other. Bots and view farms are highly unlikely to demonstrate the connectivity that human social networks do. Because of this, YouTube can freeze the view count of any video suspected of non-human activity which makes SMMs significantly less effective.