The code examples for the TPR undecidability series were intended as thought problems. Typically they were meant to fail in thought-provoking ways, ways that were usually unexpected. Truly not the kind of code you'd like to maintain.

In one of my TPR drafts, I said that, if you read my code without understanding its objective, you'd think the proof was that I have no idea how to write code. I think that remark wound up on the cutting room floor.