Lambda the Ultimate

In contrast to his earlier attempts, which I appreciated, this time Felten does not try to prove his point by explaining the fundamental facts of computer science. Understading programs is hard, he demonstrates here, because the results of some mathematical functions are hard to predict (in his example, a secure hash function).

While this enables Felten to give a short and easy to understand example, I think he misses the chance to explain why understanding (and predicting) program behaviour is hard in general. This would bring him back to the fundamental principles he discussed before, stemming from the halting problem. By going this route it would be possible to relate the issue under discussion to topics that are more directly related to LtU: how programming languages can help make analyzing program behaviour easier, how languages can restrict possible behaviours etc.

Be that as it may, more people outside the profession should be come to appreciate that understanding program behaviour is hard. This is something that takes time to appreciate fully, even when you are a programmer. It is one thing to know that finding bugs is hard in practice and quite another to appreciate the reasons why it is inherently so.