In population protocols, n different agents, which one can think of as tiny mobile computers with limited memory, run an algorithm to try and solve a particular problem, using pairwise interactions.
Consider the relative majority problem: each agent is given an opinion among k many, the agents have to compute the opinion that is the most represented.
In this talk, we will run an algorithm that solves the relative majority problem: I will ask each of you to play the role of an agent, and we will run the algorithm collectively and chaotically. Expect to move around, interact with each other, and not do a lot of sitting down. This will show how stable this surprising algorithm is to different parameters, including human error.