[cadillion]

Agreed, forgive me for reposting from another comment, but here's our perspective:

You are correct, for this smart contract to work you need defined rules around how the money is released should either party falsely claim success or failure, so you need to rely on automation or some third party to validate fulfillment of the contract. So you're back to an oracle problem, how do I know I can trust the validator?

The nice advancement here is that anyone can be that third party! Your smart contract can explicitly identify a third party for multisig, or you can rely on a validator pool of mechanical turks instead of a licensed escrow agent, and accordingly the cost of escrow can decrease.

The nice part is that this is customizable, and built into the very rails that move money. It's a much lower barrier to entry than trying to build something like that for ACH, and that's what excites us about cryptocurrency.

[Grustaf]

The thing is that smart contracts don’t solve the hard part, they solve the easy part, moving the money.

In well functioning countries that’s not even an issue at all, physical possession of the money is not very relevant. The important part is to determine who owns them.

Of course, if there is a third party that both sides trust, then you can indeed use smart contracts, but that’s exactly the situation where you don’t need them.

I’m sure someone can come up with niche use cases where so-called smart contracts have a substantial advantage, but they will not revolutionize finance.

[cadillion]

Let's take a real world example: doing a 1031 exchange to get favorable tax treatment on a home sale.

Let's say that you have a property you want to sell, there's a buyer for your home, and there's a seller of a second property you want to exchange into. In order to lock the escrow on the property you want to buy, you first need to lock escrow with the buyer for your home, and prove the sale so that the bank will give you the mortgage for the rest of the second home.

This is a fairly complex transaction that happens all the time, and the solution is to try and wrangle as many as six parties (eg property holders, lien holders, lenders, and buyers) into communicating with each other to make this happen. It often takes months!

Now imagine that you have NFTs for each property that represents the title, the banks have a flash loan contract that validates all the prices for the NFT's using Chainlink or similar as an oracle, and there's an escrow contract that validates every leg of the contract will execute an atomic swap, or none at all. People do similar activities today, all the time in yield farming. If this was the status quo, such complicated transactions could be concluded in a day, across any jurisdiction that was willing to respect an NFT swap as a transfer of ownership.

There are definitely some gotchas here to work out, as we see all the time in cases like IRON and TITAN, but we believe that these are growing pains instead of permanent unsolvable show-stoppers. And we want to be ready for the day that such transactions are commonplace and approved by sovereign jurisdictions.