Hacker News new | ask | show | jobs
by str33t_punk 2022 days ago
You would love Eth 2.0 then! The idea of proof of stake is to remove waste from the equation
2 comments

Isn't the basic idea of ETH still "distributed computer" a.k.a. "pointless redoing every computation over and over" albeit with 2.0 it's sharded (i.e. only a subset of verification nodes redoing each computation)?

Still sounds wasteful, except slightly less massively so.

That's accurate, although Eth 2 is planning to use fairly small shards (128 nodes according to the current spec). So there should be ~128 nodes redoing each computation, vs several thousand in Eth 1.

Some newer projects (including the one I work on, mirprotocol.org) are using cryptographic proofs to verify each computation, to avoid re-execution entirely.

Proof of stake does not have the same security properties as proof of work. The arguments are too complicated for me to usefully communicate them in a short HN post, but as a heuristic just keep in mind that most cryptocurrencies haven't adopted PoS despite it facially seeming better/"less wasteful".
Given that ETH is the second biggest cryptocurrency it will be a great stress test of the security properties of proof of stake.

Honestly if it runs successfully for multiple staking periods I don't think there will be much of an argument for PoW cryptocurrencies that aren't the dominate coin in their respective type of work. Essentially all coins that share a proof of work with a larger coin are constantly at risk of a 51% attack. We've seen it before with bitcoin forks.

Proof of work's security properties are broken for everything that isn't BTC, ETH (although this will change soon) and whatever the dominate CPU coin is (Monero?)

> Proof of work's security properties are broken for everything that isn't BTC, ETH (although this will change soon) and whatever the dominate CPU coin is

PoW secures any coin that dominates the daily issuance of all coins (efficiently) using the same hardware. So the different hardware classes are

CPUs

GPUs

SHA256 ASICs

scrypt ASICs

Equihash ASICs

X11 ASICs

etc...

and each one can secure one dominant coin.

> most cryptocurrencies haven't adopted PoS despite it facially seeming better/"less wasteful"

Most post-launch cryptocurrencies were designed a long time ago, though. If you look at recent credible projects, most involve proof of stake: NEAR, Solana, Coda (Mina), Cosmos, Celo, Polkadot, etc.

I can only think of a couple recent blockchain designs based on proof of work: Grin and Iron Fish.

You're right, but that's only the case right now. Incredible effort has been and continues to be made on making PoS secure, scalable, and distributed.

I'm not aware of any insurmountable problems having been found. With Eth2.0 steadily approaching launch and other networks in various stages of progression, it seems increasingly unlikely we'll find one.

The insurmountable problem is that a user cannot tell which is the 'real' fork of a PoS chain without some kind of sidechannel information.
My favorite argument: "it's too complicated for me to explain!"
See this comment by Gregory Maxwell (a core Bitcoin developer): https://news.ycombinator.com/item?id=25007874

Saying that PoS offers zero security seems a bit extreme but there's definitely a trade-off.