Considering the Person in the Puzzle: Challenging common assumptions about Sudoku player strategies

Abstract

Pen & paper puzzle games are an extremely popular pastime, often enjoyed by demographics normally not considered to be gamers. There has been extensive research into generating and efficiently solving digital pen and paper puzzle games, often by creating an artificial player. However, there have been few academic studies focusing on players themselves. We conducted a qualitative study where we observed the Sudoku solving strategies of 31 participants. Our findings reveal interesting discrepancies between common assumptions about players’ Sudoku solving strategies made by both guides and AI Sudoku systems, and their actual approach. For example, in contrast to approaching Sudokus in a systematic way and applying simple deductions—a strategy commonly assumed by AI systems—we found that a range and combination of strategies are applied to even the simplest Sudokus. Our findings suggest new directions for designers (both human and AI) of Sudoku and other puzzles, informed by players rather than models.