I think you're fundamentally correct, but with the caveat that trust inside the system can be the basis for trust outside the system via Zahavian signalling and/or escrow. As a simple dumb example, if an institution sends a million dollars to a burn address then I can probably trust that institution to handle half a million dollars in assets at a time without screwing over their customers (if they didn't intend to act fairly and make back that money over time through sustainable business practices they wouldn't have burned it in the first place). It's trickier than that -- you have to consider whether this is a sunk cost in a given context, if all customers are on the same page about what assets the institution is handling, etc. -- but the basic idea is there. Escrow is more clever than a burn pile, but could lead to some weird situations where one side attempts to extort the other for most of the escrow's value, and simply making both sides pony up doesn't obviously solve this (though it goes a long way).

Sure, but you have to trust less humans. You only need to trust the source, without needing to trust the middle-men.

So in case of e.g. a fridge the source would be some copper mine in Africa and the other end be Amazon Logistics that delivers the fridge to your house? When you draw the line somewhere else the issue to be solved it balancing transparency, trust, willingness of the channel master to share information and the tendency for powerful entities to form.

> Everything else trying to tie a physical asset to a chain has the same problem

Agree that “blockchain for supply chain” is inherently flawed. But there are hypothetical (if currently unfeasible) use cases. For example, derive a cryptographic key from a carcass’ DNA. Now you have a good physical-digital nexus. Or, for electronics, measure some random and hard-to-replicate analog jitter.