It's the government trying to enforce their opinion of who should own those Bitcoins, thereby taking power away from the owner that the network has decided on, which would be "whoever has the cryptographic keys".
A person can believe whatever they want, but when push comes to shove, it's a country's court of law that ultimately determines who legally owns what.
I think you've answered your own question - a true crypto believer does not agree with that. If the smart contract says the Ethereum is mine because you wrote it poorly and I called the transfer money function in the right way ("exploited it"), a true believer would say "yep, it's yours."
Only because human language leaves a lot of room for interpretation. Computer output doesn't, or at the very least not nearly to the same extent. If your smart contract is itself legal (you are legally allowed to formalize those terms), and produced an output as a function of it's actual internal operation (and not a random, accidental bit flip) then it should stand even in front of a judge.
There are contract disputes all the time over what a word or phrase means, and what a judge will look at is which interpretation best aligns with the broad strokes of what the parties were agreeing to. Nobody agrees to a contract that contains "I get to void the entire agreement at my discretion, keep the proceeds, and leave you with nothing"
Talking about (smart) contracts in general, if both parties agree that there was an error they can resolve it without any court. The problem is when only one party disagrees.
Imagine you have a contract with the bank and agree to pay 10% interest on a loan. Later on you try to claim you just weren't paying attention and thought it's 1.0%. That's a hell of a case to prove in front of a judge. And if that were the case the concept of contract would be worthless, invalidated by simply claiming "I didn't mean that".
A smart-contract should be easily reproducible. If that piece of code consistently returns the same result under the specified conditions then it's valid even if the result was because of a mistake the author made while preparing it.