Networks are all around us, many hiding in plain sight. They’re used to model interactions between galaxies, explain gene expressions in humans, and chart relationships via social media. But artist and data scientist Kirell Benzi is using data visualization to chart a more fantastical type of network: the popularity of this summer’s viral sensation, Pokémon GO.
Benzi calls his Pokémon Network the world’s first digital archive of Pokémon GO fervor. By charting the rise (and fall) of the game’s popularity, Benzi aims to unearth what makes something go viral, and conversely, the forces that drive online supernovas into obscurity.
Benzi, a 28-year-old PhD student at the Ecole Polytechnique Federale de Lausanne in Switzerland, created the network by scraping YouTube playlists containing the words “Pokémon” or “Pokémon GO” to catalog nearly 50,000 videos posted online since the app debuted. “The idea was to study the virality of videos to create a network, see how the network develops, and then once we have it, [study how it develops] over time,” Benzi tells The Creators Project.
The result is a dazzling interactive graphic mapping the nature of online hype, and more simply, charting the popularity of the most-downloaded game in smartphone history. Fittingly, the resulting visualization is shaped like a Poké Ball. “The shape is a bit on purpose, but it’s not all my fault,” Benzi says. “The algorithm I chose works on a force-directed layout, so it mimics what you would find in nature with the repulsion of atoms.”
In this particular mathematical structure, the most popular YouTube videos, or nodes, are pushed to the center of the graphic. Here, videos with titles like “Pokémon GO Fails Funny” and “RAREST Pokémon EVER SEEN” abound. Things get a little weirder on the outer edges of the network, where more obscure Pokémon videos are relegated.
Understanding the math behind the Pokémon Network makes playing with it exponentially more fun. Double-clicking nodes lets users pull up and watch every single Pokémon video currently hosted on YouTube, while understanding exactly where it fits into the virtual Pokémon GO landscape. From this, he compiled his own YouTube playlists of most-viewed Pokémon videos, most-liked Pokémon videos, and most-“appreciated” Pokémon videos (determined by the ratio of video likes to dislikes).
The system does not currently pull data in real time, but Benzi has taken two digital snapshots to date—once, at peak-hype mid-August, and again in early September. “One of my main aims starting out was to see the evolution of the network. Each time, we use the same algorithm to scrape data and what we witness, is that of course, the topology has changed,” Benzi says.
Unsurprisingly, his data revealed that interest in Pokémon content, in general, spiked in conjunction with the app’s release. In the first 15 days that the app was live, Benzi catalogued 30% of the total number of views on anything Pokémon-related during the entirety of YouTube’s existence. What’s more, though the media declared Pokémon GO passé, Benzi’s data proves that hype continues to grow, just at a slower rate.'Most liked Pokémon videos (03-09-2016),' via Kirell Benzi on YouTube
As the fervor over Pokémon GO lulls, Benzi plans to update the network once more in January 2017—this time, to chart the app’s spiral into irrelevancy. Benzi hopes his data can shed some light on the fickle and increasingly fast-paced rate at which people adopt and drop trends, and possibly, how to avoid letting ideas fade into the online abyss. “This is the whole point of this study. Like, how can we find the secret recipe to virality?” he says.'Most appreciated Pokémon videos (03-09-2016),' via Kirell Benzi on YouTube
Click here to visit Kirell Benzi's Pokémon Network.