That is to say they're imperfect and require human intervention to uphold intent when abstractions are incorrectly implemented over lower levels.
After all, otherwise we'll never truly be able to build very autonomous networks due to humans halting them all the time to reset some part of them.
In fact, I even think that should be the default for blockchains, letting the code be canonical, the authoritive source.
The better solution IMHO is to allow people to opt in to legal authority by declaring that a given human contract is the canonical declaration of intent, with the code declared as a tool built to help enforce it.
This way you also drop the uncertainty and ambiguity in meaning of the code and in most of the marketing, when you always have a primary document declaring what it is you're dealing with.
This will literally never happen, as it's the prerogative of the state to decide that circumstances that flatly override contracts - for example, bankruptcy (and specific things that are and aren't enforceable in regards to it), or nullifying of contracts by minors.