Luckily, there are already mitigations for risk of loss, notably backups and multi-signature wallets. Both have tradeoffs and additional mitigations. Backups are additional single points of failure but can be protected by passphrases. Multi-signature is more complicated but has no single point of failure and redundancy.
Future developments like covenants (restricting future spending of a coin) will create even more secure storage, systems that allow you to claw back stolen funds using backup keys for example.
> systems that allow you to claw back stolen funds using backup keys for example
Doesn't that remove the trust in the transaction? Isn't the one of the promised joys of these things that, short of a fork, the transaction is immutable and that people don't have to worry about fraud in the form of charge backs or their money disappearing once they have it?
If you know the mechanism for potential reversal of the transaction and trust the process responsible for making the decision about when to reverse a transaction, you've known the rules and the context from the start.
It would have to be a well-designed process to earn your trust, but it's not an unsolvable problem.
To my knowledge, there aren't any smart contracts that have implemented such a system really well yet, but I might just not have heard about them, or they may just not have been tried yet.
Well, I'd be interest in how this turns out - the only way I can envision it is either both parties need to agree to a return (which can already be done by agreement), one party has to agree (taking agency away from the other), or a "higher power" needs to agree (removing the freedom from governance that these technologies propose to offer). In most of these cases there's self-determination being lost.
The third party has to demonstrate trustworthiness, but isn't that always how arbitration of a disagreement works? Even in the existing monetary system with all its clawbacks, someone has to decide whether you're actually eligible to get your money back.
At least in this system, you could have whole services that build a track record based on publicly identifiable information. You can't flood them with reviews to skew the results because you can cryptographically verify whether the reviewer was party to the disputed transaction.
There are a thousand theoretical variants on the same basic idea, but it's at least no worse than the existing system.
It doesn't, because the systems I'm talking about don't require trust, they're built from the Bitcoin scripting language (well, proposed changes to it).
Future developments like covenants (restricting future spending of a coin) will create even more secure storage, systems that allow you to claw back stolen funds using backup keys for example.