> What's a fair threshold?

Thresholds are never fair and always arbitrary. However that's not a reason to avoid anything that has an arbitrary threshold. Arbitrary is often better than nothing.

When it comes to money, arbitrary rules are definitely worse than nothing. If I use a system to own or transfer money, I want to know exactly what the rules are. I have no confidence in arbitrary protections. Lack of confidence is what kills economies, companies, and currencies. It’s in every cryptocurrency project’s interest that people who hold the currency have confidence that their rights are protected.

Exactly, the cryptocurrency told us the code is law, but in the ETH, the Lord can change the law whenever he wish. It looks absurd.

Separate from the question of whether people should be made whole, I think, is the more interesting question of who has an obligation to make people whole.

Consider the difference between insurance companies, and reinsurance companies. Insurance companies help individuals when the individual has something bad happen to them (ignore, for the moment, that the individual has to pay into this scheme.) Reinsurance companies, meanwhile, operate solely to insure insurance companies: they keep insurance companies from collapsing when a natural disaster causes a correlated risk event that would otherwise cause an insurance companies’ clients’ claims upon them to bankrupt them.

I think the Ethereum developers-cum-NetOps are a sensible group to serve as Ethereum’s reinsurance mechanism, fixing problems for entire classes of people in the event of an economic “natural disaster.” In this case, the proposed EIP effectively reinsures this multisig-wallet “bank.”

But I don’t think it would make sense for the Ethereum devs to be made responsible for individual claims, any more than the US Mint is responsible for making you whole if someone steals your wallet with cash in it. (Though the US Mint, surprisingly, does attempt to make individuals whole if their cash is accidentally burned or shredded or melted or what-have-you, which is an intriguing basis for a counterpoint.)

Anyway, the usual party responsible for making depositors whole in the event of a system screw-up, is the bank itself. This is, in fact, effectively the reason that there are separate (consumer deposit) banks, despite it being entirely possible to just nationalize them all as part of the extended infrastructure of the central bank. These institutions are taking on the (uncorrelated) risk of holding these depositors’ funds themselves, in exchange for being allowed to themselves profit from the arrangement.

Ethereum, I think, would do well building up an infrastructure of smart contracts that facilitate bank-like obligations, including the obligation of the depositor to be made whole in the event of the bank being hacked or making a mistake.

The Ethereum devs would still be the ones to act if a mistake or accident event caused structural damage to the Ethereum economy as a whole. But, under the expectation (in such a world) that plain ETH is like cash in your pocket—uninsured—and only ETH in a [some-hypothetical-ERC-obeying] bank contract is deposit-insured in the event of contract error, the Ethereum devs wouldn’t be expected to act in such cases. It’d be your fault for choosing to give your money to a non-insured bank. (Or maybe some regulatory agency’s fault for allowing the bad bank to exist or to advertise.)

> It’d be your fault for choosing to give your money to a non-insured bank. (Or maybe some regulatory agency’s fault for allowing the bad bank to exist or to advertise.)

To fail to protect people is to invite regulation. And because people in different countries own Ether, it might happen that different national governments enact different, contradictory regulatory schemes. It would really be better for the Ethereum devs if governments didn't see a need to do that.