Dr. Blum is also a great professor for smaller undergraduate classes. I took his "theoretical cryptography" class, which was a semi-random walk through the history of creating and breaking crypto-systems. It was a small class (8-12 people), and his whiteboard lessons were great. That said, it took some getting used to the environment. The first problemset he assigned in that class started with the question "1. Think about the scientific method"; the "no right answer, but many wrong answers" approach was typical of most of his assignments. His extremely open-ended approach to learning was a breath of fresh air compared to other undergrad CS classes which seemed geared for memorization not understanding.
Multiple times throughout that course, the class collaboratively created a cryptosystem. Dr. Blum would inevitably see the cracks in them instantly, and then teach the principles so we could rigorously show weakness. None of our cryptosystems were any good, being a handful of novice undergraduates, but _collaboratively_ tackling hard problems in small pieces definitely helped me in my career afterwards.