In 2009, GitHub launched a contest to improve its repository recommendation system by inviting participants to develop programs that predict user preferences based on a dataset of repository watches. The challenge offered unique prizes, including a bottle of Pappy Van Winkle and a large GitHub account for life, to incentivize participation and innovation in open-source contribution. The contest aimed to solve the problem of recommending projects by exploring collaborative filtering techniques, particularly focusing on implicit binary interactions rather than explicit rating systems. Participants were required to submit their results via a GitHub project with a specific post-receive hook and release their source code under an OSI-compatible license. The winner was determined based on the highest score achieved by accurately predicting withheld repository watches by August 30, 2009.