On the efficacy of board game strategy development as a first-year CS project

Abstract

We report on a study comparing an open-ended freshman-level CS2 project with a fully specified project of similar difficulty. We employed a randomized, controlled trial methodology. The students needed to use similar data structures and algorithms, presented during lectures, for both projects. The first half of the milestones were almost identical for both groups, but the open-ended project asked students to develop a player strategy for a specific board game, while in the other project the students executed the game by communicating with the player modules and verifying the moves against the rules of the game. In the final stages of the project while one group of students worked on honing their own strategy, the other students implemented a pre-specified tie breaking mechanism used when two player strategies tie in a match. At the end of the term, both groups had a friendly tournament: the player strategies faced each other for the ultimate winner while the other group's submissions were ranked by a team of judges based on the implementation's correctness and style. The tournament was just for fun - it did not contribute to the final grade. We hypothesized that the player group would enjoy the project more, would score higher, and would be better motivated and equipped to study computer science. Our study confirmed higher enjoyment and interest of the study group participants. We did not observe statistically significant differences among other measured quantities, for example grade distribution, of the respective groups.