This group of experiments considers how robust ICET is under a variety of conditions. Each section considers a different variation on the operating environment of ICET. The ICET algorithm itself is not modified. In Section 4.2.1, we alter the environment by labelling all tests as immediate. In Section 4.2.2, we do not recognize shared costs, so there is no discount for a group of tests with a common cost. In Section 4.2.3, we experiment with complex classification cost matrices, where different types of errors have different costs. In Section 4.2.4, we examine what happens when ICET is trained with a certain penalty for misclassification errors, then tested with a different penalty. In all four experiments, we find that ICET continues to perform well.