Matt Levine covers this in his newsletter today:

> The other risk, which we talk about less, and which is sort of less intellectually interesting, is that the bank might lose track of the money. You might go to the bank and deposit $100, and the bank might write down “$100” next to your name in its notebook, and then it might spill coffee on the notebook and be unable to read the entry and forget that it owes you the $100. And then you might come back to the bank in a week and ask for your $100 back and the bank might say “who are you? what $100?” And the bank might be totally solvent and have invested your $100 in very safe things, but it won’t give it back to you because it doesn’t have a record of you.

> This risk is also heavily regulated, though you hear about it less. That regulation is just less controversial. There are material and debatable trade-offs in bank capital and liquidity regulation: A bank with tons of capital will be safer, but it will also have less capacity to lend to the real economy, so there are debates about what the optimal level of capital should be. Whereas with keeping track of the money, there are fewer trade-offs. Banks should entirely keep track of the money. Nobody is lobbying for “actually bank ledgers should be a bit more loosey-goosey.” The proper amount of losing track of the money is none.

> And this risk is harder to fix after the fact. If a bank loses all your money, the FDIC can give you your money back, because the FDIC is the government and can print money. If the bank loses its list of who has the money, what can the FDIC do? You can go to the FDIC and say “that bank owes me $100,” but anyone can say that, whether or not it is true. The definitive list of who the bank owes money is kept by the bank. Unless it isn’t. If the bank doesn’t keep a definitive list, then nobody does.

Funnily enough, blockchains fix this. They are ledgers were data is very hard, or impossible, to lose.