Law Smells - Defining and Detecting Problematic Patterns in Legal Drafting

Building on the computer science concept of code smells, we initiate study law i.e., patterns in legal texts that pose threats to comprehensibility and maintainability law. With five intuitive smells as running examples — namely, duplicated phrase, long element, large reference tree, ambiguous syntax, natural language obsession develop a comprehensive smell taxonomy. This taxonomy classifies by when they can be detected, which aspects relate to, how discovered. We introduce text-based graph-based methods identify instances confirming their utility practice using United States Code test case. Our work demonstrates ideas from software engineering leveraged assess improve quality code, thus drawing attention an understudied area intersection highlighting potential computational drafting.