Let me use this occasion to pose a question that might even be serious research material. Can one come up with a good measure of similarity between successive exams? That is to say, suppose you wanted to show me with suitably convincing rigor that I was really deviating from previous years. How would you do it? Of course, this would require a good theory combining formal logic, linguistics, epistemology, and of course, mathematics, both in the form of the theory and in the specific construction of the models that the theory tries to work with. I’m sure that a truly satisfactory theory would be close to impossible at this point. But it still might be fun to start thinking about it.
Very roughly speaking, there should be
a space of linear algebra exams
possibly of very high dimension, endowed with a natural inner product, using which we can measure the distance between exams. In this space, you might attempt to show that the exams of previous years form a pretty tight cluster, while my exam is convincingly distant from that cluster.
If you’re interested in thinking about a problem of that sort, let me know. I’m not an applied mathematician, but I think we might be able to do something.
