We explore how constraints act as essential scaffolding rather than enemies, examining Andy Bhattacharyya’s case for treating compliance as a design partner and Jim Lewis and Jeff Sauro’s findings that AI complexity often degrades usability. As our validation tools become unreliable, we confront the tension between deterministic heuristics and unpredictable AI outputs, questioning whether synthetic users can truly replace human preference. By pulling legal requirements upstream and admitting the limits of AI simulation, we aim to find rigor in the very unpredictability that currently challenges our design processes.
