At the end of Section 2.4.1 we find that conflicting results can be obtained from our model and suspect a problem in the decision procedure. Our difficulty is to find out where this occurs. To help us investigate, we construct a revised interpreter which tests for limited forms of contradiction. The advantage to this form of extension to our meta-interpreter is that it makes explicit the method by which we are testing our hypothesis (that conflicts in decisions cause inconsistencies in our model) and automates part of the testing of that hypothesis.