[Cryptography] threat models, was quantum computers & crypto

[Jerry Leichter]

> Many people assume the account numbers on a check are secret-ish, but
> of course they are not.  Everyone to whom you have ever written a check
> has a copy of them.  Banks' security model for checks depends on auditing
> and reversing bogus transactions, so even if someone were to use the 
> account info to make a fake check, or more likely an unauthorized ACH
> withdrawal, you tell the bank when you get your statement and they
> reverse it.
This is a mild version of what Paul Karger used to call the bull-in-the-china-shop approach to security:  You always buy replacement china, but the bull is dead meat.

Getting into the details:  Reversibility is an essential property of the entire financial system.  It's really is what makes it all work:  The system can act very quickly on an immense number of transactions because the tiny number that will turn out to be wrong in some way can be reversed on a much, much slower time scale, a time scale that allows close examination of what's actually supposed to happen.  The issues that hit the press are exactly the ones that *can't* be reversed:  The accidental huge bets on the market that go wrong (the transaction can be reversed, but at the prices in effect at the time of reversal, often good enough but sometimes those prices have moved very rapidly); or the deliberately obfuscated moves through a chain of anonymous accounts in countries that don't cooperate with everyone else.  The focus on irreversibility and anonymity by so much of the crypto-currency community guarantees that what they produce will always be marginal:  It's just too dangerous to the participants.
                                                        -- Jerry