Legacy System Trap

The pattern where a system that was once fit for purpose becomes an increasingly expensive liability but cannot be replaced because too many other systems depend on it. The mechanism: (1) the system is built and works well, (2) other systems integrate with it, creating dependencies, (3) the system ages and its technology becomes obsolete, but the dependencies remain, (4) maintaining the legacy system costs more each year (scarce expertise, outdated technology, fragile architecture), (5) replacing it would require modifying every dependent system simultaneously, making the replacement cost astronomical, (6) the organization is trapped: too expensive to maintain, too expensive to replace. COBOL systems in banking (estimated 95% of ATM transactions still run on COBOL), NASA's use of 1970s-era flight computers well into the 2000s, and millions of organizations running business-critical processes on ancient software all demonstrate this trap. The trap tightens over time: as the system ages, the replacement cost grows faster than the maintenance cost, making the economics of replacement never favorable.

When to use it

When maintaining old systems consumes disproportionate resources. When expertise for critical systems is concentrated in retiring employees. When integration constraints prevent adopting better solutions. During technology strategy planning.

How it can help

Provides a framework for preventing and escaping the legacy trap. Prevention: (1) design systems with replacement in mind from day one—modular architecture, clean interfaces, documentation, (2) set 'technology sunset dates' when systems are deployed and budget for replacement, (3) maintain the expertise to operate AND replace the system—don't let knowledge atrophy. Escape: (1) use the 'strangler fig' pattern: build new components alongside legacy ones and gradually redirect traffic, rather than attempting a complete replacement, (2) identify the minimum viable replacement scope—which dependencies can be broken first? (3) treat the replacement as a multi-year migration, not a single-event cutover. The economics: the longer you wait, the worse the ratio. Replace early.