Opening strategies in the game of go from feudalism to superhuman AI.

Abstract

How does information infrastructure shape long-term cultural evolution? Using over four centuries of professional game records from the game of Go, this study explores how strategic dynamics in opening moves reflect historical shifts in the 'infostructure' of skilled Go players. Drawing from recent work on how population size, AI, and information technology affect cultural evolution and innovation dynamics, I analyze over 118,000 games using measures of cultural diversity, divergence, and player network composition. The results show distinct eras of collective innovation and homogenization, including an early 20th-century explosion of novel opening strategies, a Cold-War-era die-off, and a recent increase in evolutionary tempo with the arrival of the internet and superhuman AI programmes like AlphaGo. Player population size shows an inverse-U relationship with opening move diversity, and a recent decline in strategic diversity has accompanied a shift in the player network, from many small subgroups to a few large ones. Surprisingly, the influence of AI has produced only a modest, short-lived disruption in the distribution of opening moves, suggesting convergence between humans and AI and incremental rather than revolutionary cultural change.