Human computation games (HCGs) utilise human intelligence for the purpose of problem solving through games. Recently, HCGs have increased in popularity within various application areas. A proper system of describing such games is necessary in order to obtain a better understanding of the current body of work and identify new opportunities for future research directions. To date, however, there is little research conducted in understanding how such games could be classified in a systematic manner. Past classification schemes rely either on quality or entertainment aspects of these games. In order to have a deeper understanding of such games, this study argues that they should be classified via different dimensions. As a starting point, this paper reviews existing literature on HCGs and then develops a typology consisting of 12 dimensions and related strategies. This typology provides a holistic view of HCGs, and provides a common understanding of the domain among researchers and designers. The typology will help researchers to better understand the nature and consequences of HCGs, and designers better identify strengths and weaknesses of different strategies for each dimension, thus facilitating the creation of entertaining and effective HCGs.