>Well, everyone has their own bank account. There's a lot of potential there to just transact directly

The technical problems with that are much less important than the legal problems.

It's likely that the IRS will maul users (unless they report every transaction as a tax event!), and the bank may refuse transactions. Users may even ask the bank to refuse transactions, and then your collateral isn't really a collateral.

Sounds technically interesting. However, it seems that they can't accept credit cards and transactions take some time, so I guess that most users will end up flocking to centralized exchanges for a better experience.

> can work trustless with oracles

Isn't that a contradiction in terms?

The amount of trust will be limited to trusting that the bank is showing your balance correctly and that nobody stole their SSL keys.

Or, more likely than stealing their SSL keys, found a “vulnerability” that caused whatever string the smart contract is looking for to appear in a signed request from the server. I put vulnerability in quotes because it's not clear to me that that is not something banks would consider part of their threat model.

It's kind of like how SMS messages worked fine until “if I can read an SMS sent to your number I can withdraw from your account” became part of the threat model.