Hacker News Reader

Comment by dozzie | Hacker News Reader

> * Distributed consensus schemes are not useful only for monetary applications.

OK, but blockchain is not a distributed consensus protocol. It's a document timestamping protocol.

Yes, a document timestamping protocol that we reach a consensus on in a distributed fashion. I'm not sure why a document timestamping protocol can't also be a distributed consensus protocol?

I do kind of find it annoying that we use the word distributed when its a clustered/consensus system. In my mind distributed should mean actors can work independently of each other but that's not really the case here. But this ship has sailed.

I'm not sure why a clustered system can't be distributed? One can always argue over taxonomy, but clustered systems are typically considered to be a type of a distributed system.

I'm also super curious why you think blockchain systems don't have actors that work independently of each other? I was always under the impression that was the entire point.

>clustered systems are typically considered to be a type of a distributed system.

Sort of. However, a Perforce server with local replicas is not the same thing as a fully offline git transaction, for example.

>I'm also super curious why you think blockchain systems don't have actors that work independently of each other?

They work as part of the network but two users can't make a transaction without having to involve the network.

True, but in practice "blockchain" implies some sort of distributed consensus.

Except for private blockchains, which are totally brain-dead.

Github does pretty damn well selling private blockchains and whitelabel private blockchain services.

Clever. I like it.

Can you explain? My understanding is we're all working on the same hard math problem, once a certain number of us have verified the answer (ie distributed consensus) one of us is given a reward to encourage further problem solving.

This is very much not what distributed consensus is. You don't "work on the same hard math problem" and you don't get "a reward to encourage further problem solving". Consensus is agreeing on a value among (almost) all protocol participants, and "distributed consensus" is proper technical term, not a vague definition by hand waving like "object oriented programming". Read about Byzantine Generals problem first.

Could one look at it more like viewstamped replication rather than traditional consensus?

Thanks for everything except the condescension!

Dude, you come to a discussion without knowing the basic concepts. What exactly did you expect?

Nope, any bitcoin client can verify any transaction very easily. (Bootstrapping by verifying every transaction in the blockchain does take a while though.) Proof of work is only to prevent a 51% attack where someone spins up enough bots to hijack "consensus" and rewrite transaction history. That's why the difficulty can be dialed up or down - most of the work in mining is not related to verifying the transactions.

Yes. People often misunderstand that verifying transactions with blockchain is intrinsically difficult which is not the case. The hard part is making a ledger that can't be rewritten.

And mining is actually less akin to solving math problems and more akin to, well, actually mining. That is, digging around enough until you find some gold.

Reply on news.ycombinator.com