This isn’t a formal paper so much as a documented think-through: a reflective write-up born from running into “legacy” everywhere in practice and needing to get precise about it. In mainstream IT, “legacy” often means out-of-support, fragile, and risky... Tidy enough. In transport and other long-lived, safety-critical domains, the word does messier work: it can signal certification lock-in, ecosystem decay, or invisible dependencies that only surface when a single tool, driver, or specialist retires. What follows is my attempt to pin that slipperiness down, not with grand theory, but with grounded observations from audits, assurance cases, and migration projects across rail, automotive, maritime, aviation, energy, and cyber-physical systems.