Visualizing Coevolution with CIAO Plots

Abstract

In a previous article, we introduced a number of visualization techniques that we had developed for monitoring the dynamics of artificial competitive coevolutionary systems. One of these techniques involves evaluating the performance of an individual from the current population in a series of trials against opponents from all previous generations, and visualizing the results as a 2D grid of shaded cells or pixels: qualitative patterns in the shading can indicate different classes of coevolutionary dynamics. As this technique involves pitting a current individual against ancestral opponents, we referred to the visualizations as CIAO plots. Since then, a number of other authors studying the dynamics of competitive coevolutionary systems have used CIAO plots or close derivatives to help illuminate the dynamics of their systems, and it has become something of a de facto standard visualization technique. In this very brief article we summarize the rationale for CIAO plots, explain the method of constructing a CIAO plot, and review important recent results that identify significant limitations of this technique.