Hacker News Reader

Comment by yxhuvud | Hacker News Reader

No, the fundamental assumption is that distributed consensus in itself is useful. So far, there are some applications (databases), but for the stuff they are using the blockchain for, not so much.

View on HN

If distributed trustless consensus is useful, there are cheaper ways to get it than PoW mining. It's just that most folks don't understand them.

The principle value of Bitcoin as a protocol is its incredible (even detrimental, from a technical perspective) simplicity.

Are there other means of distributed trustless consensus as secure as PoW? Secure being the operative word here. Lots of attempts, but none yet convincingly as or more secure.

People who are not crypto-anarchists don't realize that the only reason for proof of work is so you don't have to trust anyone. If there is trust, a centralized database is better in every way vs. Blockchain.

People who do not read modern research don't realize lots of the things they think are true about the state of research are totally false.

They also think that it's Proof of Work that provides a lack of trust. While that's involved, what's actually more meaningful is the demand to quickly select a "profitable" branch to mine and re-mine on. That can exist in PoS as well.

Nakamoto consensus isn't the only protocol that exists; there's also things like Raft and Paxos with different tradeoffs.

What you mentioned does not cover byzantine faults... pow is the only known solution

There is a large amount of r&d being put into BFT-style proof of stake for Ethereum, it’s always been part of the roadmap. Much work still needs to be done beyond the test net, but so far no fundamental showstoppers have been found. There is also constant collaboration between the researchers and developers of major projects to move the entire field forward.

It’s still early days, unfortunately the hype got far ahead of the tech but progress is being made everyday. I think the same of VR/AR.

https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQ

If only someone could think to google "BFT Raft extension" or "BFT paxos extension" and find published research. And if only papers from that search had citations. And if only those cited had links to foundational research on generalized BFT strategies. Oh if only we lived in such a world.

If only. If Only. IF ONLY. iF oNLY. If OnLy. iF oNlY.

Of course, but those aren't trustless per my question. They're designed for single administrative environments like corporate data centers where a trusted entity controls the environment.

The other two possible environments being multi-administrative (permissioned DLT deployed among a consortium of banks or entities sharing administrative privileges) or no-administrative (public permissionless blockchains like Bitcoin/Ethereum). The hard problem is developing consensus algos for the latter that scale while maintaining security of the shared ledger.

Hedera Hashgraph recently revealed that it supports more than a million transactions per second based on current usage by developers, that's enough to support decentralized servers and open source MMO environments. It uses gossip protocol and a form of voting protocol.

1) Non-technical friendly overview: https://www.youtube.com/watch?v=MzWiiOLv96I

2) https://www.hederahashgraph.com

I'm invested a bit in Credits which claims it'll do a million transactions per sec but it's kind of bs - they can do nearly half a million in very limited conditions but in the real world the numbers will probably be far less and I'm a bit skeptical of such claims. I'm not sure that fast is necessary though - credit cards do a few thousand transactions a sec globally so if you can handle that you could potentially at last use crypto to pay for your coffee etc https://medium.com/@viruseslovers45/credits-the-best-blockch...

Yeah, Hashgraph appears to do high tps at relatively small number of nodes, aka will probably work in permissioned/consortium style environment with limited nodes, but whether it can scale in a public environment with unlimited nodes is an open question.

You only need to beat like 14 transactions a second to start doubling and tripling bitcoin's average throughput.

A non sharded public blockchain that doesn't do some kind of validator delegation will always be bottlenecked by a single validators computational capacity, which is on the order of 100-1000. Hedera is really pushing the line on making fraudulent claims here.

Hashgraph doesn't use blockchain - look into interviews with Dr Baird if you're interested. MachineZone, the big game company, is now building on Hashgraph:

https://www.prnewswire.com/news-releases/hedera-hashgraph-co...

I am aware of their architecture. How does it circumvent the bound?

Could you provide a couple sources please, I believe many of us are not well read on blockchain literature.

This is a good primer [1].

[1] https://softwareengineeringdaily.com/2018/03/26/consensus-sy...

Reply on news.ycombinator.com