Black Mirror Bandersnatch: data mining your decisions
After reading many comments from friends on Facebook about the new interactive show from Black Mirror called “Bandersnatch”, I decided to watch it last night. This is not the first time watching Black Mirror, as a matter of fact, I love it! This reminds me of one of my favorite old TV shows: Twilight Zone. I still get scared when I watch the episode: “Nightmare at 20,000 Feet”, when the gremlin is on the airplane wing! (The gremlin in the movie is scarier than the one from the old TV show though).
I don’t want to make spoilers about Bandersnatch, so I won’t tell you the endings and how did it go in my experience.
Some interesting facts about this movie:
- Took more than a year to plan
- The script is 157 pages long! (Black Mirror scripts are usually around 65 pages)
- Filmed over 35 days
- You can explore over a trillion permutations but only five possible main endings
- The script was made using Twine
This is very interesting, the script was written using Twine, an open-source tool for interactive fiction that has been used to both create and prototype video games, but this may be it’s most high-profile use of formats outside of traditional gaming.
You can check it out by yourself!
Twine / An open-source tool for telling interactive, nonlinear stories
You don't need to write any code to create a simple story with Twine, but you can extend your stories with variables…
Following the story and Data Mining your decisions
The story is about Stefan, an 80s programmer adapting the fantasy novel “Bandersnatch” into a video game. It has five possible main endings, depending on the choices you made during the interactive show.
If you have watched it already, you can see that at some point, Stefan, the main character becomes aware that there’s somebody controlling his decisions.
While making my decisions through the show, some of them trivial and others increasingly horrific in Stefan’s life, I thought how clever from Netflix to be data mining decisions from all the Netflix users, such picking Frosties or Sugar Puffs for breakfast and others quite scary!
Is there an important purpose for collecting this kind of data?
Netflix got another interesting way to gather user data through this interactive show that invites you to make countless decisions and this could be added as another method to monitor us in an attempt to understand our view habits. I’m not surprised that Netflix is a True Big Data Company. Big Data Analytics is pure gold for recommendation engines, which are so popular for companies such as Amazon, Pinterest, and Best Buy.
So, what is Netflix doing with the user data it gathers from “Bandersnatch”?
Just think about all the possibilities…from choosing products like cereal to choosing your creepy decisions along the story, I’m worried about the last point, hahaha. Netflix could exploit and learn more about us and our preferences.
Playing with the Data!
I collected data from different movie reviews websites and found Black Mirror “Bandersnatch” reviews. Played a bit with the data and used my favorite data visualization Python library: wordcloud. I also used the symbol from “White Bear” episode to give it a creepier look! Got these interesting results:
Here, I divided the user reviews into good and bad reviews. Least than 3 is a bad review and more or equal to 9 is a good review. You can see in the wordcloud the most used words in the comments, the big words are the most popular. This is just a sentiment analysis to analyze people reaction to the TV show.
I also applied sentiment analysis to the titles. Users are very creative!
And of course the dates! You can see that most people gave their reviews just right away! The show was released on December 28, 2018.
This new idea from Netflix blows my mind but also makes me think about the data mining behind this. Are we going to see more product adds on social media? Frosties or Sugar Puffs?
Let me know your thoughts!