Why is Unit-testing in Computer Games Difficult?

In this paper, we investigate how unit testing can be used to verify computer game simulations with implicit and poorly defined mathematical models. The difficulty of calculation of expected values for such cases tends to lead to Fragile and Obscure tests. The effects of these test smells can be reduced if we improve the strategy we use to calculate expected values. We investigate four strategies and while each of these is insufficient, we outline how they can be combined into a single, sufficient strategy.