Hacker News Reader

Comment by dboreham | Hacker News Reader

Thought experiment: if you were tasked with creating an eventually consistent distributed database where the nodes were to be run by different organizations that didn't trust each other, with predefined validity constraints on the data of some sort, and a scheme for incentivizing the operators of said database to keep operating it long term, how would you propose doing that?

Hint: if you look at similar systems such as those used for certificate transparency logs, they look quite like...a blockchain.

View on HN

Thus, a problem in search of a solution. Who has asked for that exactly ?`It's been more than a decade and still no one has found an actually useful real-world application for blockchain besides gambling and money laundering.

Now that’s the 1m dollar question. Virtually no one

I'm just a newb passing, but a self sustained decentralized 24/7 logic+network layer could be nice for stupid computational tasks like tracking physical items or updating multi-source (different services, different companies) data easily, which are often done at the human level today (even if 99% of the time it's not necessary and very wasteful). Again i'm not knowledgeable but when thinking of blockchains, I always have this in mind.

It can't track physical items, because you can always just lie about the data. Company A says "I put the jewels in box 1352", and Company B opens the box to receive a bunch of dish towels. Just because it's cryptographically verified, doesn't mean it has any semantic value.

If Company A and B already trust each other not to put bad data on the chain, then they also trust each other to just send emails back and forth, and you don't need a cryptographically verified blockchain. It secures the least important part of the process.

But how would you do that with ~all companies ? it would be standardization and heterogeneity hell.

Why would the addition of blockchains help standardize things?

they standardize transport, verification, encryption at least, consensus too iiuc

No, they don't do any of that. There are 100 different blockchain implementations at least, and none of them are standardized across any of those metrics. Everything from the consensus mechanism to the packet format to the bytecode is different. The standard way to interop between the Tezos blockchain and the Polygon blockchain is to create a token on both sides and attach a tag saying "don't touch me I'm actually somewhere else".

The only way to standardize ~all companies would be to have them all run the same exact ledger system, and if it was easy enough to make them do that, it would be easy enough to have them standardize a different reporting system for the same purpose.

i meant each but using a single one to host events and logic, not a multichain (yeah according to your input would be brittle at best)

IBM tried to do that (asset tracking for marine shipping) with some rather neat technology.

It sank like Titanic. Nobody was interested in it, turns out that companies are fine with emailing each other Excel sheets.

Yeah I think i remember stories about that, that said, a failed attempt doesn't mean future will not be different (nobody seeing programming languages evolution as comparison point). Neither will it means that a decentralized event chain store is the solution.. just saying it kinda felt like an interesting "small logic global network application layer" .. which felt way more efficient than the litany of different b2b apps.

Sure, but that’s not a thing any actual business needs, except in rare, esoteric situations. For all actual business purposes, a centralized database is better fit for purpose (note that pretty much all finance runs, eventually, on centralized databases like the Fed’s ACH system and the Depository Trust Company. It’s fine.)

You cannot reply with a thought experiment to someone asking what real purpose a technology has. Provide a real example where this is needed/done

Tasked by whom? In pursuit of what goal?

Reply on news.ycombinator.com