> Consider the example of origin / provenance tracking. That's not a problem that needs to be solved by blockchain unless the company can't trust its own audit logs and database team.
It is if:
* you want if you want more than one company to cooperate
* you want participation to be permissionless
* you want it outlive the life of the particular corporations involved.
* you want join ownership
* It is also desirable if you think they could be hacked, or have downtime--ie, to remove the company as a single point of failure.
A bitcoin-style blockchain outsources trust to the participants with the most computing power. It does not magically eliminate trust.
This means that it's only a good idea if either a) you can limit the participants, and you somehow trust the participants with the most computing power than (e.g.) a simple majority of authorized participants, or b) you do not limit the participants, and you trust anyone in the world with the most computing power more than you trust anyone you would pick for some other reason.
For some applications - e.g. money for people who don't trust their governments or aren't well-served by the inter-government norms around currency exchange - they're legitimately in situation B. Sending and receiving bits to literally anyone is better than not being able to send and receive currency at all, and since a 51% attack will be noticed, it's unlikely to be used on any individual small-time user.
For provenance tracking between a set of companies identified in advance, I have difficulty understanding how you can be in situation B, and you're not in situation A.
If you instead want a cryptographically-verified immutable append-only log, there are lots of non-blockchain structures for immutable append-only logs that permit only a finite set of writers. Certificate Transparency is one of them: issued certificates cannot leave the log, and there are mechanisms for pre-logging certificates before issuance, but the only things worth logging are certificates from trusted (or partly-trusted) certificate authorities. Anyone can participate in the trust process, but transactions themselves only come from a limited set of participants, so you get to completely bypass proof-of-having-more-computers-than-everyone-else.
If you instead want a cryptographically-verified immutable append-only log, there are lots of non-blockchain structures for immutable append-only logs that permit only a finite set of writers
Yes. I actually like the concept of Bitcoin, but it's very annoying how so many of its fans seem to think Satoshi Nakamoto invented Merkle trees.
I'm actually okay with using "blockchain" to mean "distributed Merkle tree" if that's what people want. Just be clear about what properties you want so nobody builds mining into your system when unnecessary.
Hold up. That's really reductive. It isn't just a merkle tree. It's a merkle tree that you can look at and say "this merkle tree would cost 2 billion dollars to forge." This is valuable because if someone hands you one, you can tell if you're looking at a fake.
I didn't say Bitcoin is just a Merkle tree. My point is that Bitcoin is a system that combines different pre-existing components - like Merkle trees and Proof-of-Work - in a novel way, but people then split that system, and call those components "blockchain". Some user here in HN actually wrote "Git is a blockchain".
But the best example is that nonsense of "permissioned blockchains".
icebraining's claim is that Satoshi did invent something, but the thing he invented was not Merkle trees.
If you have an application where forging or forking the tree isn't meaningful (e.g., any possible Certificate Transparency Merkle tree can be merged at any point into any other one), then the thing that Satoshi invented is not useful to you, and having it cost $2 billion to build such a Merkle tree is a waste.
I don't really understand how a blockchain allows participation to be permissionless (how do you know what's written there corresponds to reality?) so maybe that's where the magic is, but every other point has been solved for decades - the DTCC does essentially what you're talking about for stock ownership.
It is if:
* you want if you want more than one company to cooperate
* you want participation to be permissionless
* you want it outlive the life of the particular corporations involved.
* you want join ownership
* It is also desirable if you think they could be hacked, or have downtime--ie, to remove the company as a single point of failure.