Hacker News Reader

Comment by vertex-four | Hacker News Reader

Unfortunately, a pile of tech companies decided to start selling "blockchain" to banks which doesn't contain any distributed consensus system at all, never mind a Sybil-resistant one. As a result, people in the finance world and people surrounding that particular subset of the tech industry have a very warped idea of what a blockchain is.

Under the definition often used in finance, all modern Linux machines are powered by blockchain, because journald uses hash-linked entries to provide evidence when log entries are manipulated.

View on HN

To be fair the term "blockchain" is too broad and generic to give it a very specific definition. I think that's a big reason for the hype, like "cloud computing". It's so vague that nobody really knows what it's about, if you actually decide to get technical you lose 95% of the audience as soon as you utter the words "cryptographic hash" or "public key cryptography". So you can just make any bold claims without going into details and still be sort of right. I can say something like "blockchain technology is going to be big for avocado cultivators to be able to audit production year after year". Does it make any sense? Who knows. But it keeps people going.

And it's probably true too. Blockchain technology is here to stay. Cryptographic hashes, public key cryptography, merkle trees, hashcash, peer-to-peer networks, all that is here to stay. Of course, all that existed before the Blockchain, in the same way that servers existed before the "Cloud" but it doesn't matter, it's all about that sweet branding.

As a really quick hint, if somebody pushes blockchains as a solution for anything but cryptocurrency and managing cryptocurrency based on the state of other cryptocurrency, or a solution for Zooko's Triangle, they're selling a lie.

EDIT: the reason for this is that if the chain ever needs to touch something that isn't entirely stored within the chain - for example, the number of avocados picked on a certain day - it needs a trusted oracle to tell it that. If you have a trusted oracle, you can trust that oracle to host a bog-standard SQL database and avoid this whole mess, unless your boss is breathing down your neck about using "blockchain" so your company can look cool and hip to its investors/clients.

What about a registry of land ownership?

That could be tracked fully on a blockchain with the added advantage of no longer needing to trust some centralized office to 1) correctly track ownership and 2) properly store their records.

A similar idea exists for tracking ownership of domain names.

There are other cases where you trust the oracle but want some form of non-equivocation. A nice example is a notary. In that case, just an SQL database doesn't suffice. Sometimes, something like a pgp-signed statement could work for that.

If you want some more resilience against the claim 'that was signed with a fake key' you could mandate all signatures be published on an actual block-chain. This time-stamps them so you can see whether they happened before a key compromise, and allows the requirement that the counter-party audits the blockchain to ensure that no signatures are published that they did not know of. This audit would mean that signatures published with a compromised key would quickly be detected.

This way, the records of the notary are publicaly checkable. You now only need to trust the notary at the time of publishing their signature on the blockchain (with some extra time to ensure they can't claim to not have audited yet).

Public records does seem like a reasonable use case. Birth, death, marriage, ownership etc. Not sure it's a lot better than what we have now but it might be cheaper. (Or the reverse.)

With public records, a big question is whether you need some form of distributed Sybil resistant consensus.

> What about a registry of land ownership?

Towards what purpose? So that a court can look at it and decide who owns the land? In which case the court can run a SQL database for significantly cheaper than a secure proof-of-work system.

The notary idea is a good one, except that it's not actually a problem that needs solving in most countries with a semi-functioning democracy. If you want to timestamp something relatively important, you post it in your city/county/country's newspaper of record. In the UK, for example, that'd be The London/Edinburgh/Belfast Gazette. Various entities maintain copies of these, in case you're worried that the Government decided to edit their copy of the record.

Consider title insurance [1]. The current system of land registry in the U.S. is so disjointed that you need insurance against the registry making mistakes. This might be solved by making a centralized database, but then we are depending on the technical capabilities and security of the court system. Just because we trust the court to make correct decisions does not mean we trust the court to keep that data correctly and publicly available.

Besides, a system like this could allow for land transfers that are recorded only in the blockchain, without need of an intermediary.

A harder objection would be complicated deeds. For example, a deed that grants a piece of land but requires that, when it is sold, the proceeds go to some charity. For cases like that, you'd require a signature by a trusted third party (an oracle). We are now trusting that oracle only to make the right decision when the signature is requested. Moreover, it is now public data that there is a complex deed behind this transaction. Thus it is no longer possible that upon sale, the deed has been forgotten and then later it turns up wreaks retroactive havoc.

Another issue would be 'property loss'. People lose their crypto keys all the time, what happens when people lose the 'key' to their land?

[1] https://en.wikipedia.org/wiki/Title_insurance

I won't claim to be an expert on this, but isn't that the whole point of multiple / decentralized oracles?

Reply on news.ycombinator.com