Hacker News Reader

Comment by fshbbdssbbgdd | Hacker News Reader

I think this article is saying that proof of stake doesn’t work because the community can decide to ignore the consensus algorithm and pick a different, modified chain. Proof of work has the same property though, as exhibited by the author’s own example of ETH vs ETC.

View on HN

I think the author's point is that in a PoW fork, the hash power is a fixed resource and must split between forks. In a PoS fork, prefork holders can stake the same amount of coins on both forks, thereby "doubling" the total coin being staked. With PoW, the fact that the hash power has to split between forks gives a mechanism for a consensus between forks to be found. With PoS forks, you might as well just continue to stake both forks.

There's already a solution to this: https://eth.wiki/concepts/proof-of-stake-faqs#what-is-the-no...

>This can be solved via two strategies. The first, described in broad terms under the name “Slasher” here and developed further by Iddo Bentov here, involves penalizing validators if they simultaneously create blocks on multiple chains, by means of including proof of misbehavior (ie. two conflicting signed block headers) into the blockchain as a later point in time at which point the malfeasant validator’s deposit is deducted appropriately. [...] The second strategy is to simply punish validators for creating blocks on the wrong chain.

Which one does ethereum use?

I believe Casper (Ethereum's PoW protocol) uses the second strategy.

> That is, if there are two competing chains, A and B, then if a validator creates a block on B, they get a reward of +R on B, but the block header can be included into A

What if it can't (we do ETH 2.342 variant (which doesnt cost us anything contrary to PoW forks) with modified header and later it wins)?

On PoW, if one fork is much less popular, it will be feasible to attack because you can rent enough hash power to produce a larger chain. If only slightly less popular, this isn’t feasible.

On PoS, a substantially less popular fork will be possible to attack, because you can afford to buy enough coin to stake and take control of the validators. The economic value of your coins will be divided between the forks based on market consensus. A more popular fork will be hard to attack because the coin will be too expensive.

In my view these are a similar type of security guarantee, what am I missing?

In both PoW and PoS, providing security to prevent attacks requires allocation of economic resources (spending hash power, or holding and staking tokens).

In PoW my GPU can only hash one chain at a time so I have to pick a fork to continue on. This is true for everyone else too, so the incentive is to mine the fork that will most likely live since you can’t earn coin without mining.

In PoS, I hold tokens and when a fork happens, I’d hold tokens in both forks now (same parent token in both child forks). I can trivially run 2 validators where I stake my tokens in both forks. This is possible because validation in PoS was meant to not consume much resources, so running 2 is affordable. The incentive is to stake both forks because doing so means I’ll pick the right fork (by picking all) and thus preserve my stake in at least one fork.

Not all validators will run their node on the fork. As seen with the 2016 fork, many will sell their forked token to buy more of the real thing.

I think the punishment system described elsewhere in this thread solves the issue of existing validators running both chains. If that’s correct you’d have to buy more coins to stake to “attack” the cheaper chain (which is much like renting hash power to attack a PoW chain).

But in practice, hash power isn't used to choose between forks. Nobody says ETC is invalid because it has fewer hashes than ETH. It's still a valid chain, even though it has less market value. The same would be true of PoS: one fork would have more legitimacy and market value than the other.

If you're not actually forking, but just operating the protocol normally, PoW chooses between branches just fine, but so does PoS. With Ethereum's version, there's even "finality," which means it's impossible to revert more than a few minutes without destroying a third of the total stake: https://ralexstokes.medium.com/the-finality-gadget-2bf608529...

That's a strict improvement over PoW which only offers a probabilistic guarantee. Both systems can be bypassed by a hard fork, but in both cases the fork will face the same challenge of market value and legitimacy.

> Nobody says ETC is invalid because it has fewer hashes than ETH.

I think you are quite wrong there. At some point the exchanges had to pick which chain was going to get the ETH ticker and which chain was going to get the ETC ticker. While it's hard to say exactly why they did what they did, certainly the amount of hashpower on each chain was a significant factor.

I would add that some proof of work blockchains are non-forkable e.g. polkadot

I think the difference here is that doing so would be a hardfork and would require everyone to opt into the change. Contrast this with PoS where if 51% of the stakers woke up one morning and decided to rewrite the last 2 weeks of the chain they can immediately do so, and anyone who disagrees must opt out.

With slashing PoS, if stakers tried to rewrite the left 2 weeks of the chain they could have their stake taken away as a penalty for signing votes on multiple forks.

According to this other comment[1], it looks like as long as you're on the winning side you're not punished?

[1] https://news.ycombinator.com/item?id=27237610

And both chains continue - but it usually becomes very clear which chain is stronger - and the weaker chain can be killed off because it can be attacked more easily - which devales the value of that chain. 24 vs 520 in this case is the has rate difference, or a 20x difference.

I guess... how long does it take to determine this? I have a transaction on a PoS PoS (lol) chain, and then... I guess... I've given my goods away for some coins. Now... it turns out my coins are lost because it was on the weaker chain?

Not trying to be disingenuous here. Is this occurring in a matter of a few blocks, a few days? Does this bring crypto closer or further from potentially being an actual currency one day?

I tend to think this centralizes things (PoS as opposed to PoW). I can't imagine I'd want to put my money into some whale controlled currency personally. But I'm not sure that part has really changed considering I think of PoW as relatively whale controlled (and rather centralized, and against all of the generally accepted Keynesian economic theories).

A POS POS chain may not have a ton of value or attacks. But when splitting off, ETC for example probably was hit by some attacks until the value was gone.

The way it works is whoever accepted your coins can have them lost and you could respend them if you were the attacker.

https://www.coindesk.com/ethereum-classic-blockchain-subject...

Reply on news.ycombinator.com