Title: An experimental study of the generation of inequality and unpredictability in an artificial academic literature market
Abstract: Academic publishing has changed substantially in the past 30 years due to the advent of the internet. Unlike print publications, digital publications have the ability to provide additional information to visitors about the publication and previous readers via metadata like download counts. This study investigated the effect of this metadata on the development of download inequality and unpredictability of success in an experimental academic literature marketplace. We found that presence of an accurate download count increased inequality in article downloads, meaning fewer papers accumulated a larger share of the total download count. We also found that the presence of download count increased the unpredictability of success, meaning across identical instances, different papers became the most popular. Finally, an exploratory analysis found papers were more highly rated when download counts were present. Together, these findings provide insight into how the download behaviors of previous academic readers may influence literature choice.