Obscure documentation and critical process knowledge scattered across inaccessible systems make replacing an incumbent look expensive and risky, so keeping them looks like the safe choice.
An example
Fifteen years of work by one outside contractor had gone into a hospital system's billing platform, and the chief information officer had been told for three years that bringing it in-house would save four hundred thousand dollars a year. Every replacement plan stalled. Business rules lived in stored procedures with no version control, and variable names changed from module to module. The documentation the contractor delivered each quarter was undated, scattered across three file sites, and missing the exceptions that made month-end work. The in-house billing team, asked to write runbooks, said they were trying but the system was complicated, while their daily output stayed perfectly normal. When the chief information officer ran a trial cutover at one facility, the month-end close failed and a payer report missed two required fields. The chief financial officer asked who had approved the trial. The contractor was renewed for three years.
Why it works
Buried knowledge makes keeping the incumbent the easy path: no transfer effort, no migration, no risk. It makes replacement look worse, since rebuilding what is buried eats much of the projected saving. Above all it makes replacement costly and uncertain, with failed closes, regulatory gaps, and an executive who owns the wreckage. Decision-makers discount what they cannot measure, so they treat the buried complexity as larger than it is. Long tenure and autonomy over systems and documents build it, helped by a principal who cannot tell real complexity from manufactured obscurity and by the absence of documentation or rotation standards. Teams resisting standardization get the same effect by leaving critical processes undocumented.
The thinking behind it
Oliver E. Williamson The Economic Institutions of Capitalism (1985) shows that relationship-specific assets, such as knowledge of one firm's systems, leave the other party dependent and exposed to being taken advantage of. An incumbent who buries process knowledge raises that specificity on purpose, so replacing them looks ruinous. Third-party inventories, pair coverage, and shadow runs reduce the specificity, which is why they break the hold-up.
George A. Akerlof The Market for "Lemons": Quality Uncertainty and the Market Mechanism (1970) shows how a party that cannot judge quality is left to price what it cannot see, and the better-informed side gains from the gap. A principal who cannot tell how complex the buried systems really are must assume the worst, and the incumbent profits from that uncertainty. Costing the cleanup turns the unknown into a known, removing the premium the incumbent was collecting.
Warning signs
Critical processes run from one person's laptop or one team's memory. Documents are undated and spread across several sites. Documentation deliverables slip repeatedly while operational output stays normal, with the explanation that the system is complicated. Every replacement plan comes back with a migration cost that swallows the savings. The same name answers every question about how something really works. Trial cutovers fail at exactly the steps nobody wrote down.
How to defend yourself
Arguing the merits of replacement does not move this. Turn the unknown into a costed project.
- Commission a third-party audit that inventories the buried systems and prices the cleanup. A bounded estimate removes the fear premium.
- Require pair coverage, a second person who can run each critical process, before any change of contract or staff.
- Make documentation a sign-off deliverable, artifact by artifact, tested by having someone else run the process from it.
- Schedule shadow runs during quiet periods, with successors executing the process while the incumbents watch.
Accept some short-term operational pain, and warn the people who will feel it. Once the worst case is a costed project, the hold-up loses its force.
How a leader shuts it down
Prevent the trap at the start, with contracts that carry documentation obligations, source-code escrow, and transition assistance. Set runbook standards, and rotate people and require pair coverage in critical roles. Audit documentation by testing whether someone else can use it. Give the executive who leads a transition the time and backing to absorb a rough quarter, and do not punish a first failed trial that exposes what was buried.