The Exam Boycott That Taught Me About Cooperation And The Future Of AI

In the fall of 1988 as a first year at Hamilton College, I took a class taught by Dan Chambliss that featured a unique game theory experiment. The professor challenged us with a final exam proposition: if no one took the exam, everyone would receive an A, but if even one student showed up, anyone who stayed away would get a zero. I took on the mission of organizing a boycott, utilizing printed documents, a music stand, and personal outreach to build trust. Ultimately, the boycott succeeded, and no one took the exam. Nearly four decades later, I reflect on this lesson in collective action, comparing the challenges of human coordination and trust to the future problem of getting autonomous artificial intelligence agents to work together.

What a college class the first semester of my first year in college in 1988 taught me about leadership and collective action. Also: why getting intelligent people to work together can be harder than making them intelligent in the first place.

In the fall of 1988, I arrived at Hamilton College in Clinton, New York, and signed the college register in a small clapboard building at the foot of College Hill. I was officially matriculating at an institution with origins stretching back nearly two centuries—and with a founding vision that was itself an experiment in bringing people together.

I had no idea that one of the most memorable experiences of my college years would involve a professor, a final exam, a student-led boycott. And little did I know that all of this prefigured a game theory problem that would eventually become relevant to the future of artificial intelligence.

I enrolled in a sociology class taught by Dan Chambliss. What unfolded over the semester, in that class, became an extraordinary experiment in organizing, trust, incentives, and collective action.

A College Founded on an Ambitious Idea

That spirit of institution-building appears throughout Hamilton’s history and among its alumni.