My point there was that the key aspect that makes something a smartcontract is that it automatically causes the property transfer when certain triggers are met, in a sense that is just as real as in the system's normal property-transfer primitives.
Escrow, as it has conventionally existed, tries to accomplish that same thing, but isn't quite that thing. In the moment, the escrow agent must themself manually initiate the "triggered" property transfers, and is capable of violating the rules, hence why they have to be a trustworthy party. Sure, they can be sued later, but in the moment they can effect the transfer.
This is substantively different from a smartcontract where that all happens automatically.
Escrow, as it has conventionally existed, tries to accomplish that same thing, but isn't quite that thing. In the moment, the escrow agent must themself manually initiate the "triggered" property transfers, and is capable of violating the rules, hence why they have to be a trustworthy party. Sure, they can be sued later, but in the moment they can effect the transfer.
This is substantively different from a smartcontract where that all happens automatically.