> That argument holds some ground - think of all the armored cars, shipping containers, and facilities that serve to move and store cash and gold, as well as all the labor involved in maintaining the system.

There are very good counterarguments to that as well.

The idea that the "banking system" functions primarily by shipping cash and gold around is nonsense. 99.9% of traditional banking nowadays happens on computer networks. The Visa network for example handles 150 million transactions per day (that is, 500x more than bitcoin) while using a tiny fraction of the electricity. When a bank in London wants to send money to a bank in New York, they don't ship a container full of gold, they send the money electronically using the SWIFT network. To the extent that cash is still physically shipped around, it's only because people still sometimes use cash to pay for things, and banks collect and distribute this cash and replace damaged bills.

And your estimate of labor is similarly flawed. To begin with, the "banking system" includes an enormous amount of services that Bitcoin doesn't provide (loans, mortgages, securities), and retail banks are part of that as well. The transactional parts of banking can be (and are being) largely replaced with mobile apps and ATMs. The other aforementioned parts of banking aren't replaced by Bitcoin, so all of that labor still needs to exist.

So the efficiency argument shouldn't be "what if BTC replaced banks", it should be "what if BTC replaced SWIFT, ACH, BACS, Visa network, etc." there is a value proposition here but it is much smaller than the one originally proposed.