Legacy system

A legacy system is not merely an old system. It is a system that persists not because it is optimal but because the \'\'switching cost\'\' — the cost of migration, retraining, data conversion, and organizational adaptation — exceeds the perceived benefit of replacement. Legacy status is therefore not a technical property but an economic and sociological one: a system becomes legacy when the organization around it has invested more in working around its limitations than in understanding them.

The danger of legacy systems is not their age but their \'\'invisibility\'\'. They become infrastructure — the water in which the organization swims. Decisions are made assuming their constraints are natural laws. The legacy payroll system that cannot handle contractors shapes hiring policy. The legacy database that cannot represent many-to-many relationships shapes product design. The legacy email system shapes communication norms. The system\'s limitations become the organization\'s limitations, and after enough time, nobody can distinguish between the two.

Legacy systems are also epistemic fossils: they encode assumptions about the world that were once true and may no longer be. A legacy banking system built on the assumption of batch overnight processing encodes a world without real-time transactions. A legacy logistics system built on regional warehouses encodes a world without global supply chains. These assumptions are not documented; they are \'\'implicit\'\', embedded in data structures and workflow logic. The system continues to enforce a model of reality that the reality has outgrown.

The tragedy of legacy systems is that they are not abandoned because they fail. They are abandoned because they succeed too well — they become so deeply embedded that replacement requires not technical migration but organizational reincarnation. And organizations, like people, fear death more than dysfunction.