Nobody can predict the future of how business will get done years down the road, but I think Smart Contracts can potentially utilize both Oracles (systems like Chainlink) to provide objective data about the world for triggering Smart Contract logic, and Arbitration systems (something along the lines of Kleros) for humans to make subjective judgements about human concepts like "reasonableness" in contracts.

A very contrived example. Say a Smart Contract exists for a contractor to paint a boat cornflower blue and then get paid when the job is done, but the Oracle system says that the contractor messed up and painted it sky blue. It can be part of the Smart Contract in the event of some error that it can go to arbitration to some pre-agreed-to subject matter experts that both sides agreed to beforehand for a quick and reasonable resolution or fee.

This kind of thing can be planned for and made standard in smart contracts. It'll only happen if it saves businesses more time and money than it costs though, and many other question marks will be involved. I think it's very exciting conceptually though.

But if a bug allows you to delete the contract. It’s useless. In the real world, a contract exists unless all parties agrees it doesn’t. A “loophole” won’t cause money to be removed from escrow unilaterally.

Actually didn’t a judge just rule Citibank wasn’t entitled to get its money back after a mistake like that?

https://www.forbes.com/sites/jonathanponciano/2021/02/16/cit...

Citi accidentally paid back lenders, rather than accidentally sending money to some unrelated party. This doesn't seem relevant.

A very major part of that decision was because the payment was not reasonably assumed to be a mistake. Obviously that argument isn't going to work here.