Although I still generally use standard problem sets in my theory classes, I find myself experimenting more and more with alternatives, including more open-ended assignments. One big motivation for this change is to put some science back in my computer science class, by making assignments more like laboratory exercises in chemistry and biology: students have to experiment to find an answer. I also find that open-ended assignments offers more room for student creativity, although it is not clear to me that students prefer this. (Indeed, it is clear that some students resent it.) Or perhaps, more pessimistically, it is just my fear that the solutions of all the standard problems are available somewhere on the Web. In an effort to encourage a conversation in the community about the topic, I thought I would present one of my recent assignments. The fall text of the assignment lies below. (I have made only small modifications, so please forgive its rough form.) In one sentence, the thrust of the assignment is that the students should model the game show Who Wants to be a Millionaire? and use their model to determine the best player strategy. The problem is harder than it might seem, and it raises many interesting questions about how to design a reasonable model. I gave this assignment to my graduate seminar on Probabilistic Analysis and Randomized Algorithms, a class primarily meant for first and second year graduate students in theory and networks. I belive that beginning graduate students and advanced undergraduates leaning towards graduate school should be learning how to develop models for complex systems. Designing proper abstractions is a useful skill for both theoreticians and non-theoreticians; this problem has the advantage of being easily understood by everyone. While I think the assignment is far too open-ended for an undergraduate class, a simpler variation might make an interesting project. (At the very least, underlying it all is a reasonably well-motivated dynamic programming problem.) Student opinion was mixed, with the primary complaint centering on open-endedness of the assignment; one said that they could spend an arbitrary amount of time on it without getting closer to the final solution. I thought this was an appropriate metaphor for graduate school. The next biggest complaint was that students were not familiar with the show, which is somehow disturbing and refreshing at the same time. If anyone wants to use this assi~ment or some variation in a class, you have my permission and encouragement. I would ask that you let me know how it goes and if you have any suggestions for how to improve it.