Endless Runner

During the course Gameplay Prototyping and Game Metrics I have learned various techniques and criteria with and according to games can be measured. I have applied these methods to a prototype of an endless runner to optimize it according to previously defined goals.

In the prototype the player controls an avatar that constantly moves forward on a left- and rightwards restricted track, overcoming obstacles through various actions or by evading them. In the first iteration of the prototype the following types of obstacles were implemented: Some obstacles, refered to as blocker, are static or moving objects that the player can avoid by adjusting the horizontal orientation of the avatar. Shield barriers are obstacles that are too extended to avoid them. Instead, the player has to activate a shield that allows the avatar to pass the barrier successfully. However, the shield has a cool down and thus must be used with caution. The third type of obstacles is embodied by 90° branches where the player must either run left or right at the precise moment since the track ends with a wall in one of the directions. If the player fails at one of the obstacles it results in the immediate death of the avatar and the round is over. The goal of the game is to accomplish a distance as long as possible. In order to give the player variety the track is randomly generated from predefined sections at run time each round.

The task was to optimize the prototype in regard to the highest possible generated flow value. For this purpose interested testers were asked to play the endless runner and then to fill in a questionnaire. The questions were based on the so called flow short scale. In addition, some parameters such as the track section at which the tester failed were logged into csv files. The gained insights were used to tweak the prototype. The most serious change was the replacement of the 90° branches by T-junctions where it is not obvious in advance in which direction the track continues. Instead, the player is signaled the correct direction by two arrows just before reaching the junction. With the feedback of the testers it was possible to assess the obstacles more precisely in their difficulty. By deliberately placing obstacles of various difficulty, a pacing curve ideal for the experienced flow should be achieved. Other changes were the introduction of a starting area to facilitate the start of a round and the adjustment of the controls. In a second test run the resulting version of the prototype was subjected to a control test. The optimization turned out to be successful because an increase in the experienced flow could be measured for all participants.