Most publicly available datasets are pre-cleaned, making it difficult to practice data cleaning and management skills with authentic, raw data. To bridge this gap, we developed Chaos—a web application that generates messy datasets from clean data.