Then almost by definition I can't really trust your smart contract, except for airtight formal verification, which nobody really does plus I highly doubt smart contract programming languages are formal verification languages.
So this:
> Yes, the output is guaranteed to be correct.
is provably wrong and you either know it and are malicious or you don't know it and this discussion is useless.
I think we're talking about two different things. Imagine you have a generic program A that you've formally verified. Now you ask me to execute it for you. I say: "ok, the output of the program is X." Can you trust the output? No, you cannot trust the output even though you've formally verified the program. If you formally verify the smart contract, you can trust the output.
So this:
> Yes, the output is guaranteed to be correct.
is provably wrong and you either know it and are malicious or you don't know it and this discussion is useless.