I could imagine something like this being applicable towards large legacy codebases with a history of memory leaks and segfaults. With legacy codebases you're often only able to tread water and won't be in a position to track everything to it's root cause.