Researchers evaluated how well metrics used by computer algorithms to solve Nonograms (a type of logic puzzle) match human perceptions of difficulty. They found that these metrics do not accurately predict how hard or easy humans find the puzzles. However, they did discover that experienced solvers tend to prefer more complex puzzles, which contradicts what the algorithmic metrics suggest. The study used a combination of computer algorithms and human subject testing to gather data on puzzle solving strategies. --- Why it matters: This research matters because it highlights the limitations of using automated metrics to evaluate puzzle difficulty. It also provides insights into how humans approach problem-solving, particularly in the context of complex logic puzzles like Nonograms. Source: https://arxiv.org/abs/2608.23300