A framework for biometric playtesting of games

A framework is described that can assist game developers in using biometric (psychophysiological) methods while playtesting. Biometric methods can give developers a valuable additional window on the playtester’s experience. 1. THE USE OF BIOMETRICS When developing a game (or any other media product), is important to collect as much feedback from users as possible. After all, users will determine the success or failure of the game and their decision to play, buy and recommend a game depends partially on having a smooth and satisfying play experience. Several methods for collecting data during playtests have been developed, with interviews and observaton still being the most popular ones. Newer methods include logging game metrics and collecting data about the physical state of the player, called biometrics or psychophysiology. Both game metrics and biometrics add objective measurements to the subjective results from interviews and player observation. The industry has widely recognized the role of game metrics in playtesting and in the continuous evaluation of a game after it has gone live. Game metrics are often used for marketing (conversion rate), but it feed into user experience analysis [5]. In a large number of studies over the past years, the value of the biometric approach has been demonstrated, see reviews in [1] and [2], but the methods is infrequently used in practice. We think the low rate of adoption of the biometrics method has to do with a number of factors, the most important one being lack of a widely adopted and generic framework for measuring this data. For game metrics, several commercial frameworks exist (playtomic.com, mochibot.com, flurry.com) and it is relatively easy to program a proprietary system. Although expertise is required for the successful in-depth analysis of game metrics, a number of widely accepted measures can be automatically computed by the existing frameworks (conversion rate, time on game, heatmap of player deaths, etc.). A framework has been developed for emotion extraction from mixed-media and biometrics data, but this has not caught on with the industry [3].