Factor Analyses on Positive and Negative Evaluations of Games against Go Programs

Analyzing users’ preferences is important for many platforms to further improve users’ satisfaction or make recommendations. This is the same for online game platforms. In this paper, we target a classical board game, the game of Go, and investigate the factors that make human players feel enjoyable when playing against Go programs on a website. In addition, we also investigate whether different players feel enjoyable in different ways. We use game records collected from the website, where players can evaluate games as enjoyable or not. We conduct statistical analyses using basic information, such as game lengths or players’ win rates, as well as advanced analyses using information extracted by a strong Go program, such as the qualities of moves. The results show that some factors are generally common among players, while some factors show completely opposite preferences. For example, players generally prefer opponents with proper playing skills; meanwhile, some prefer close games and some prefer to win by large margins.