[acdha]

That still doesn’t require a full blockchain: 90s PGP with multiple signatures is all you need, or going really old school, mailing a copy to yourself using registered mail.

Imagine if a notary, post office, etc. set up a signing service: relational database, basic signatures for user-submitted hashes, etc. Not quite as good but much cheaper to run and much easier to have trusted by a court — I suspect most users would see no additional advantage by adding a blockchain.

[oarsinsync]

The change is in the trust model.

The centralised approach described requires that one trusts the database administrator. For a lot of things (probably the vast majority), this is good enough.

For the rest, there's decentralised distributed immutable ledgers.

[acdha]

Trust is an interesting topic: I think many people – and especially courts – would be actively disinclined to trust an anonymous blockchain but might be interested in one where it’s identified participants with reputations, liability, etc. Pay the USPS, Western Union, etc. and get a signature backed by insurance.

Trying to explain something like Bitcoin just seems like you’d have a hard time meeting reasonable doubt standards in a dispute but that’s not the only model.

[malmsteen]

Exactly, most people have done so for centuries. Of course with 99% reliability only, but no one cares about the 1% possibly added by blockchain (with a retarded additional computing burden to deal witg).