Youtube and Information Cascade

I dont know about you guys, but I pick which youtube clips based on how many views they have. Generally, the ones with the more views are usually a lot better than the ones with very few views. However, everyone once and awhile you find a clip that is very popular but their is absolutely no reason for it popularity.

The reason is behind these clip popularity is because of information cascades: people assume their is a reason why a clip is so popular and click on it and thus make it even more popular and help mislead the public even more.

Here is an example of a clip:

https://www.youtube.com/results?search_type=&search_query=vodka+girls&aq=f

The person who posted the clip assumed it is so popular because of horny kids on youtube but is more like caused by information cascade.

It is for this very reason that I don’t surf You Tube using the amount of hits or looking at titles. All a video has to do is have a clever title to get a few interested people, and then all of the sudden the cascade starts and it’s a #1 hit video. The place where I do use cascades, however, is for downloading popular music through I-Tunes (yes, I actually pay for my downloads). I’ve had to keep up with Top-40 hits for the last few years due to DJing parties. Like it says in the book, cascades can be wrong and they can be based on very little information. This information is really just a reflection on what the record companies and promoters start in the network. Due to the cascade affect, however, it’s always pretty safe to play a song based on the number of downloads it has because although some people may not like the new hit song, they are most likely familiar with it and ready to dance and sing along.

Another place where information cascades are useful are when downloaded software like from download.com. If I'm looking for lets say a new antivirus to use, I look up reviews made by experts in how well they perform, etc. Another place I look for is at download.com where I sort the programs not by reviews made by editors or even the ratings given by users but by downloads this week. I'm guessing that if it is a good program then a lot of people who have downloaded it this week and would have an overall very high download total. I don't like sorting by user reviews because if only 5 people used the program and gave it 5 stars and the program is over 5 years old it can still show up as the top listing. By choosing downloads this week, I'm going off the fact that so many people before me have downloaded it, they must know that its a good, useful antivirus, so I follow in their path.

I also use this strategy when buying used books on amazon.com. They show you the ratings for the each seller and the number of ratings. I will usually go for a seller with a slightly lower rating if they have a considerably larger number of ratings. Like Piotr said, I get skeptical of the ratings when there are too few of them because one person could have a bias or be too picky and decrease the rating by a lot or the seller might not be that great but the few buyers were relatively pleased. In this case I figure the sellers have to be pretty serious if they have many ratings and if they have a few messed up orders along the way that's ok.

Another application of this idea is in an iTunes library, where you sort your songs by the most played. This is a self-reinforcing "rich get richer" scenario, because the songs that get played more often are near the top, thus more likely to get played even more often. However, a major difference between this example and others is that there's only person involved, so in effect you're trusting yourself and the first few choices you made the first time you began playing songs in iTunes. Any (decent) song could have become the most played, its really only chance.

Social Networking was the main source of communication in The Rescue, event raising awareness to the Invisible Children. Participants were asked to twitter, post videos on youtube asking for rescue, and write letters to the media and moguls. A good example of information cascades was the spreading of the video showing participants dancing outside of Harpo Studios. I was a participant and immediately after we recorded this song and dance, it was posted on youtube. People all around me were texting friends to spread the video and make its presence known. Within 2 hours, there were 300 views. Today, the 3rd day since it has been posted, there are 19,515 views. Oprah didn't rescue us that day, but the following day; I think this information cascade of the video was one of the reasons for her to watch this video and realize the importance of this cause.

Oprah Rescue Us