Brainstorming
For this project, I was thinking about the algorithms around us that constantly influence our worldviews and waste our times. I can’t think of a specific art piece in particular, but I really enjoyed the cornerstone idea of Dada making fun of or tearing down the status of technology and media. George Grosz’ paintings and Marcel Duchamp’s concept of “readymades” were good examples of this – depicting/using a technology in a ridiculous purpose. Back then, they had radio, TV, and motors, but today we have social media, the Internet, and these other invisible “algorithms” that influence us greatly. I thought that if I somehow “appropriated” these predatory algorithms into a game, I could get people to think more actively about how the algorithms influence them. I had an idea to borrow the idea of a Wikipedia speedrun (where players race against each other to use Wikipedia hyperlinks to jump from one random topic to another) and apply that to a social media algorithm. It seemed fascinating to use social media to get to a target endpoint while constantly battling a recommendation algorithm trying to keep you in a rabbit hole.
I chose YouTube because I felt it was a good balance between user and algorithmic control over the feed. Initially, I wanted to use a scrolling platform like Instagram or TikTok, but it was tricky to figure out what a target could be, since there’s so many niches and no public data from these apps showing what the public enjoys watching the most. It also provides very little user agency over what they see – all they can do is scroll and watch for a certain amount of time to game the algorithm’s parameters. Instead, I chose YouTube – its recommendation algorithm allows users to click on different options, allowing them some agency over what they watch, but they are also limited in their choices since the algorithm decides what they see in their feed.
At first, I was discouraged because I found out that this idea is not entirely new. Youtube streamers have done “YouTube speedruns” where they try to get to a certain video from surfing through YouTube recommended videos. However, I realized that these YouTube speedruners only tried getting to a topic that the player set for themselves. This was a limitation because inherently knowing that topic meant that the player would’ve had to have been introduced to that topic already from YouTube. I wanted players to experience trying to break out of their own recommendation algorithms into somewhere they don’t normally traverse or know about. Very fortunately, here is public data over the top searched YouTube queries on Google Trends, which I conveniently “appropriated” as well for my targets. I decided on a Bingo format, so players can strategize which YouTube topics to hit.
Playtest 1
For an initial viability test, I tried simply seeing if a player can get from one search query to another in a reasonable amount of time. I took the top YouTube queries and picked 2 at random, and tried to see if 2 players could get from one to the other. I gave people the option to use their own algorithm or start fresh if they were not comfortable, but people surprisingly were willing to use their own algorithm. So from their own algorithm, I assigned them to move to a baseball player. Neither participant knew who it was, so I gave some clues / similar queries, like the clue that he’s from baseball and which team he was on. While the playtest wasn’t long, the players were able to get pretty far from their search algorithm.
Playtest 2
In the second playtest, I tested out placing the Top 25 YouTube search queries in a 5×5 Bingo card. There were some questionable queries like “Victoria’s Secret Fashion Show 2025” that I debated leaving off the card. I ended up keeping it on to leave the queries as untampered with as possible. Even if the player is uncomfortable with the search topic, they have the flexibility to choose different topics to hit because of the Bingo mechanic. I had 3 playtesters Nysha, Joel, and Kenny. Here were each of their screens:
Joel:
Kenny:
Nysha:
Photos from the playtest:
Below was Joel’s documentation on the different YouTube topics he was able to catch. N
Unfortunately, nobody got a bingo, but it was interesting to see how each person strategized to make the algorithm give them something that they wanted. Whether it be cutting the video short to not encourage too much watch time, or rapidly clicking on new videos to trigger the algorithm to force something new. The players said they had a lot of fun trying to get out of a rabbit hole, and were surprised at how suffocating it can be sometimes if you are craving an escape. Overall, a very interesting playtest that I think successfully got people to think about the algorithms in a critical way.