A Continuous-Space Description of Video Games: A Preliminary Investigation

Research on the impact of video games on the brain and behavior typically focuses, at its root, on particular experiences inherent in different types of gameplay. Historically, genre labels have been used to quickly classify games into (at least somewhat) nonoverlapping types of experiences. Genre labels have been successfully used in previous research to describe differences in experiences across games. However, modern video games increasingly borrow conventions from myriad genres. This increased complexity of modern games calls to question the applicability of genre labels to such research endeavors. Here we sought to determine whether it would be possible to create a valid continuous description of video games instead. We crowdsourced recognizable lists of 100 comparative retro (1985–1995) and 100 modern (2013–2020) video games. Fifty thousand human triplet comparison decisions regarding those games were then used to seed a machine learning algorithm to establish a continuous two-dimensional space in which individual video games could be situated. These two game spaces (modern/retro) were found to have reasonable validity and could be used to visualize and support beliefs about the game space, and how it has changed over the past 30 years.