Information Cascades in YouTube Searching
http://www.business.uzh.ch/professorships/entrepreneurship/workshops/media/fu.pdf
http://www.youtube.com/t/press_statistics
About 72 hours of video are uploaded to YouTube every minute. Over 4 billion hours of video are watched on YouTube each month. With such mind blowing stats, YouTube is the obvious website of choice for many who seek to watch videos online for free. And with literally years of video on the site, the act of searching for video is an interesting one and displays yet another instance of information cascades.
When one types in the keywords looking for a certain video for the first time i.e. the person has never seen this video, a lot of times the search returns multiple ‘versions’ of either the same video or many videos that would qualify as a good result for the person. The question is, what video does the person choose? Now YouTube does a great job to stir you in the ‘right’ direction by displaying important attributes of the videos such as total views, video quality and uploader. This can result into an information cascade because the person is making a decision given information from previous decisions already made by others i.e. to watch the video or not. The total views in particular play a major role. Most people would choose a video with the most views simply because they feel since many people have watched it, it must be the best version of what they are looking for.
In most cases the right video is the one that has already been viewed the most. Whether it is always rational or not to choose the highest viewed video is not a given. One important point worth considering is that in information cascades people choose not to rely only on their own information but also inferences made from preceding patrons. One explanation for this is of course that the previous choices were made based on information that one does not have. But in the case of YouTube videos, when a person is choosing a video to watch out of many similar looking videos, the previous views were by people who had not yet seen the video. They, like you, did not know anything about the video when they decided to click to watch it. So imitation of the behavior of others is here resulting into a bad decision. And yet it still works more often than not.