Difficulty in videogames: an experimental validation of a formal definition

This paper synthetically presents a reliable and generic way to evaluate the difficulty of video games, and an experiment testing its accuracy and concordance with subjective assessments of difficulty. We propose a way to split the gameplay into measurable items, and to take into account the player's apprenticeship to statistically evaluate the game's difficulty. We then present the experiment, based on a standard FPS gameplay. First, we verify that our constructive approach can be applied to this gameplay. Then, we test the accuracy of our method. Finally, we compare subjective assessments of the game's difficulty, both from the designers and the players, to the values predicted by our model. Results show that a very simple version of our model can predict the probability to the player has to lose with enough accuracy to be useful as a game design tool. However, the study points out that the subjective feeling of difficulty seems to be complex, and not only based on a short term estimate of the chances of success.