Comparative API complexity analysis of two platforms for networked multiplayer games using a reference game

In this paper we propose the quantitative analysis of the complexity of a simple reference game implemented on a particular gaming platform as means for characterizing how the platform succeeds in easing the development of networked multiplayer games. We first present our own open source tool based on Sneed's Object-Point (OP) method for the automatic quantitative assessment of the complexity of a software API by analyzing a source code using the API. We then apply our tool, together with the recently released JSComplexity tool based on classical software complexity metrics, to compare two platforms for networked multiplayer games, the open source realXtend Tundra SDK and the proprietary Union. As the reference games we use existing implementations of the simple Pong game atop the two platforms. Our data shows that these complexity metrics reveal API design tradeoffs, resulting in complexity differences in the reference games.