The power law of practice has been a long-standing theory of the learning curve: how skill improves with experience. Despite the general agreement that learning fits a smooth curved pattern, there have been specific areas (piecewise learning curves, improper aggregation, and plateaus) that have refuted the existence of single smooth learning curves for a given task. The present study attempts to generalize strategy-specific learning curves to a large longitudinal dataset of games in StarCraft 2, a highly complex task with refined measures of skill. Using novel methodology to balance error across high dimensional measures, the ubiquity of the power law of practice is not supported, yet shifts in strategy do not account for the lack of power law learning curves. The existence of hobbyists (people who do not want to become experts) is considered as a possible explanation. Implications of the findings and directions for further research are discussed.
Copyright is held by the author(s).
This thesis may be printed or downloaded for non-commercial research and scholarly purposes.
Supervisor or Senior Supervisor
Thesis advisor: Blair, Mark
Member of collection