As a typical college student, I rightfully have a Netflix account, so I was interested in the article “How Netflix Reverse Engineered Hollywood” by Alexis Madrigal right away. Like the author, I had noticed the strange genres that form after watching movies and shows on my Netflix account and how specific they could be. Madrigal and her colleagues discovered that Netflix has 76,897 unique ways to describe types of movies. Netflix does not have all the movies to fit these genres, and some of the genres most likely don’t exist as movies yet, but the algorithm used by Netflix creates these thousands of genres. These genres are very accurate for the viewer and Netflix claims that this algorithm helps their member retention rate by predicting what sorts of shows viewers will watch. I decided to log on to my Netflix account and see how well this algorithm works for me.

Even before I used Netflix, I have had a specific way to choose what show or movie to watch next:

  1. Have I heard of that movie?
    1. Have my friends talked about it?
    2. Did the advertising for the film or show surround me in the forms of billboards or commercials?
  2. Did I hear good things about the movie?
    1. I would either watch the movie or show based on the opinion of a friend or
    2. Go to Rotten Tomatoes and let the professional critics decide for me.

I am a movie snob, so I won’t watch movies that get bad reviews, especially when they cost 10 dollars in the theater. My time is valuable and if I’m going to waste it binge watching on Netflix, I want it to be time well wasted. I never go off the stars Netflix users rate movies because they are always far too generous. To me, this star rating “feature” is so inaccurate to my tastes that I consider it a “bug.”

  1. Have I heard of the actors?
    1. If it has Johnny Depp in it, I’m going to watch it.
  2. Is the cover art visually appealing?
    1. This is very shallow, but if the cover looks cool, I might just watch it out of curiosity.

On my Netflix streaming page, the first “based on your interest” genre that pops up for me is “Critically acclaimed dramas”, based on my interest in Pulp Fiction and Cast Away. The movies that come up in this section are all classics, and I’ve seen a lot of them already. This is an excellent genre for myself and I applaud Netflix’s database for choosing this for me. (I even added some of these movies to my watch list to view after I submit this blog post.) So far, this database is working well for my interests, as it was designed to do so. The rest of my personalized genres are more straightforward, and I don’t have any of the wild crazy ones that are possible based on Netflix’s algorithm. However, I’m sure if I start watching more weird looking films, I could have something like “Violent Thriller About Cats For Ages 8 to 10.”

Madrigal also explained how Netflix trains large groups of people to watch movies and rate and tag them according to their system. However, sometimes Netflix doesn’t strictly follow these tags. For example, a category that comes up for me is “Because you watched Breaking Bad.” In my opinion, the options that come up demonstrate one failure of Netflix’s tagging system because it inserts shows Netflix produced itself first for advertising purposes:

 

reader

I don’t think The Adventures of Puss in Boots is related to Breaking Bad at all, so this is a useless genre and a failure of the tagging system. The Netflix VP might call this a “feature,” but I consider it a “bug.”

 

To conclude, here is a genre I found on the internet:

netflix-fruits

https://uproxx.files.wordpress.com/2013/04/netflix-fruits.jpg?w=650&h=195