Oversight that runs on institutional memory doesn't run very long.
Government programs lose their institutional memory faster than any private-sector organization. Political transitions, agency reorganizations, term-limited appointments, and ordinary attrition mean that the people who know why a program works — or doesn't — are constantly being replaced by people who have to learn it all over again. The knowledge exists in documents and meeting minutes, but nobody can find it fast enough to matter when oversight is needed.
Between administrations, entire programs get re-examined as if for the first time. Inspectors general and oversight committees ask questions that were answered years ago — because nobody can surface those answers in a form that's useful under time pressure. Oversight bodies end up reconstructing history from scattered files rather than exercising genuine accountability. By the time they've assembled the picture, the window for intervention has closed.
Programs themselves can't surface their own status without manual reports that are outdated before they're submitted. Program managers spend weeks assembling data for quarterly reviews that should take hours. When a flag does emerge — a contractor behind on deliverables, a grant recipient with anomalous expenditures, a regulatory gap nobody spotted — it surfaces through a human noticing, not a system seeing. That's not oversight. That's luck.