By Bryan T. Bayne
Humanity collectively spends billions of hours on video games each year. Most would brush off such figures as mere trivial entertainment, but Attila Szantner, a web developer, and Bernard Revaz, a physics researcher, saw in them one of the world’s greatest untapped resources. If only a tiny fraction of the time spent on video games could be devoted to science, researchers might quickly find the answers to thorny questions, they reckoned.
Enter Massively Multiplayer Online Science (MMOS). Founded in 2014 by Szantner and Revaz, the company connects video game developers to researchers who seek assistance from citizen scientists. The premise is simple: a background in science is not needed to adequately perform mundane tasks such as pattern recognition or image classification, therefore, by gamifying such tasks, the huge gaming community may contribute thousands of hours to assess large data sets, considerably speeding up scientific research. The project has garnered the attention of several universities, game developers, NGOs, and even of the European Union, which has provided over 570,000 EUR in funding.
The first match came in 2017, between the Human Protein Atlas (HPA) project, which seeks to build an image-based map that details the spatial distribution of proteins in human cells and tissue, and CCP Games, the developers of EVE Online, a massively multiplayer online game. EVE Online is a science fiction title that attracts some half a million players each month. Its player base is generally more mature and keener on science than other gaming communities, making it the perfect fit for MMOS.
The result was the mini-game “EVE Discovery,” which is seamlessly integrated into the video game narrative and players may access at any time to assist the HPA project. After completing a tutorial, players must classify hundreds of thousands of high-resolution images using 29 pre-defined terms such as “nucleus,” “cytoplasm,” and “mitochondria.” When enough players reach a consensus for the same image, it is marked as “solved” and sent back to the scientists at the HPA. Participation yields rewards such as in-game currency which players may use to purchase exclusive items.
A 2018 paper published in Nature compared the players’ performance with an AI trained to do the same task. It found that player classifications were on average slightly more accurate than AI ones. Humans outperformed computers when classifying less common protein classes, such as microtubules, whereas computers were more accurate when large amounts of training data could be provided. Notably, researchers found out that feeding player data into the AI vastly improved the machine’s performance.
In 2020, MMOS partnered the American Gut Project, a research outfit that aims to map all human gut microbe DNA sequences, to Gearbox Software, the developers of Borderlands 3, a popular, action-packed first-person shooter game. The resulting Borderlands Science mini-game is simple: the microbes’ ribosomal RNA gene sequences are disguised as colorful bricks in tiles on a grid. Each color is a different nucleotide, and players must match the bricks on the x-axis to those on the y-axis—each representing a different microbe. The resulting data allows researchers to estimate the degree of similarity between each microbe. Within the first month, over 700,000 players had participated, solving over 36 million such puzzles, and devoted more than 86 years of playtime to citizen science in exchange for in-game items.
The idea has caught on. “Gaming and gamification” has become an official European Commission project category; the Commission recommends that project proposals request approximately 1 million euros in funds. Hence, the European Union is now funding a dozen similar projects, which include novel ideas such as using video games to treat cerebral palsy or to train farmers to use new technology. These EU-funded projects usually connect diverse stakeholders such as universities – which play the roles of software developers and research institutions – think tanks, and private companies.
MMOS has proven that citizen science is an effective way to analyse large data sets. Now the game is on to find new applications to science gamification.
Picture Credits: Pexels, Francesco Ungaro