A game with a purpose (GWAP) provides a way to label images to support the work of search engines. For example, Google Image Labeler is a game where paired users win points by guessing the same label for an image. Would Google, however, be better off by deploying KissKissBan (KKB)? KKB involves a third player, the blocker, who competes against the collaborating pair. At each game turn, the blocker has time to type in blocking words. The collaborating pair suffers time penalties if it should guess a blocking word. The authors claim that KKB is fun to play and an efficient way to gather labels, and also that the labels gathered are more diverse.
A small questionnaire survey with 17 players supports the claim that KKB is indeed fun to play. In a gameplay comparison with a reimplementation of the ESP game (on which Google Image Labeler is based), KKB yielded 11.54 distinct labels per image versus 6.56 distinct labels for the usual two-player game, supporting the claim that the labels gathered are more diverse. There is, however, no evaluation of the quality of the labels. Regarding efficiency, the evaluation is again lacking, and there is no comparison of the actual time spent playing the games. A demanding reader would like to know if it is better to have two three-player games or three two-player games.
Despite these shortcomings, this paper is strongly recommended to those tackling the image-labeling problem. Competitive KKB might just be a very efficient and effective GWAP.