But if it does, you will always be able to blame things on user error so long as a user is in the loop. I'm sure the crash diagnostics output from a billion dol

But if it does, you will always be able to blame things on user error so long as a user is in the loop. I'm sure the crash diagnostics output from a billion dollar bomber is only available to the manufacturer and if it is available to an independent third party investigating a crash, it's probably only parseable/understandable by the engineers who wrote the code in the first place. If there is only place where corners definitely get cut, it's on returning diagnostic for functions/processes that didn't need to be debugged/logged in order for the product to make it not just out of the lab but into production. Production bugs are the hardest to diagnose, because they are the least likely to have their causes instrumented/measured/logged.