I think some confusion arises because that "smart contracts" only make sense if code really is law, in the sense that any transaction executed by the contract -- even unexpected, surprising transactions -- is considered to be fully consented to by all parties interacting with the contract.
I agree that that's a terrible idea - bugs can always exist, and having no recourse when millions of dollars are lost due to a coding error is a huge and unreasonable risk.
But otherwise -- if, ultimately, courts can force "smart contract" transactions to be unwound if they are found to be exploitative, unintended or otherwise invalid -- then what's the point of having smart contracts in the first place? What's the value proposition? Why not just use regular contracts?
