I don't really follow the argument. If the number of errors in the vault were nonzero, it would most likely indicate that the vault did not screen some ("real", "physical") effect that could cause errors. For instance, you could have errors due to radioisotope decay, low-probability interactions with neutrinos, etc.