Mapping the outer reaches of the learning curve: Complex intellectual skill performance after decades of extensive...

Little is known about the outer reaches of learning curves for very complex cognitive skills exercised over decades. Can skill performance improve as long as practice lasts or do all learners ultimately plateau? Furthermore, does natural talent set widely varying performance limits or do all learning curves eventually converge? Chess skill learning curves were examined for 333 players, mostly grandmasters, who, over a median 20 years, played at least 1500 internationally-rated games. Curves of nine of the 333 participants who played more than 3050 games, and one of the 333 and two others who played more than 4250 games over more than 30 years, also were examined. Players on average reached an approximate plateau by around 1200 games. This asymptotic value has changed little from the 1990s despite massive changes in the chess environment. Out to more than 3050 or 4250 games, players stayed at an approximate plateau for many games and years and then performance eventually declined. Chess skill learning curves do not rise forever. Over extensive practice lasting decades, they typically plateau for a long time and eventually decline. The more talented tend to plateau later and curves of the greater and lesser talented do not converge.