Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mainnet Merge Announcement

blog.ethereum.org

Post reply on HN
Print view View original

211–220 of 609 posts

Re: Mainnet Merge Announcement

#211

Post by kkielhofner »

What surprises me most about the Ethereum ecosystem is the software and release engineering of the implementations, especially the reference (now execution layer) go-ethereum.

As has been noted elsewhere "the merge" has been in progress for years. Here we are approximately three weeks out and just this morning the (allegedly) working merge ready go-ethereum client was released. Given that the Bellatrix upgrade is scheduled for 9/6 this gives the approximately 4500 Ethereum nodes (of which 3381 are geth)[0] 10 days to update...

It also doesn't help that they botched the prior release (two days ago) of the geth client that had a nasty corruption issue that requires a little more than the update warning footnote in this release to fix[1]. Many people who deployed 1.10.22 are re-syncing from (almost) scratch.

Even if you followed Prysm and the other Beacon implementations the deployment and configuration changes required are non-trivial[2] and again, the v3 release referenced in this announcement was only released two days ago!

As someone who runs Ethereum nodes I'm a little gun shy at this point deploying new releases of this stack, especially given how things have gone the past two days. Even with our relatively trivial application this gives little time for any testing or assurance process.

It's amazing to me given the stakes (value of the Ethereum chain) and complexity (huge) node operators are essentially in the position of waiting until mere days (or hours) to upgrade their nodes and as noted - there are thousands of us around the world.

IMO this is yet another indication that the real issue with regard to decentralization of blockchain solutions is the fact that (as pointed out by Moxie and others) the vast majority of the real users and platforms of Ethereum interact with the chain via a handful of centralized node providers (Alchemy, Infura, etc). Who can blame them given what a mess this is?

I'm also not picking on Ethereum specifically, the same things could be said about most of the other chains and implementations I've interacted with. Given the hype, promise, age, and value in the blockchain ecosystem the software itself makes Apache circa 1996 look like mature, rock solid software.

BTW, the Ethereum Foundation alone holds approximately $1.6B in assets[3] and investors have poured tens of billions of dollars in this ecosystem. IMO the quality and process of the fundamental software enabling all of this is inexcusable - it's not like they're wanting for resources to do this right.

[0] - https://ethernodes.org/

[1] - https://github.com/ethereum/go-ethereum/releases

[2] - https://docs.prylabs.network/docs/prepare-for-merge

[3] - https://www.coindesk.com/business/2022/04/19/ethereum-founda...

Re: Mainnet Merge Announcement

#212

Post by drdeca »

spywaregorilla wrote: post #120
programmarchy wrote: post #112

Earlier quoted context omitted.

The slashing can only occur if the validator can be proven to break certain rules e.g. double voting. The reports are evaluated by the network.

> The reports are evaluated by the network. The network which was just shown to be 51% malicious actors? And how is it evaluated? What if the slasher is dishonest? What if it sends trillions of fake requests?

Evaluated in the same way that “does this address have enough to send to this much to this other address” is evaluated? If it isn’t valid, then a block containing it isn’t valid, and this is a thing which any client can automatically check.

What happens if you try to send trillions of invalid transactions in any sensible chain?

Re: Mainnet Merge Announcement

#213

Post by rglullis »

MuffinFlavored wrote: post #191
bluelightning2k wrote: post #143

I can be anti-crypto and still appreciate this. First - clearly reducing the environmental impact of anything by this much is pro-humanity. (Although having the impact to begin with is another story.) Secondly from a sheer technical coordination perspective there's a feeling of pulling off a complex dance. Makes it hard for any of us to claim our workloads aren't testable!

> First - clearly reducing the environmental impact of anything by this much is pro-humanity. How long until BTC follows suit?

Judging by the BTC maxis on this thread and all the FUD they are spreading, never.

If I have to make a bet, though, I think that what will happen is that financial institutions will start pushing for the idea of wrapping BTC on the Ethereum blockchain, and once it reaches a certain threshold (let's say 80%) they will campaign to drop the bitcoin PoW altogether.

But this is a conversation that bitcoiners are not ready to have, yet.

Re: Mainnet Merge Announcement

#214

Post by cesarb »

TakeBlaster16 wrote: post #38
spywaregorilla wrote: post #10

Is it incorrect to say that ethereum is now entirely centralized with some extra steps? Like, you have this proof of stake thing, but the only reason it works is because there's just a small number of validators, which is just going to be the ethereum foundation and friends. edit: putting this at the top because nobody is responding on topic. I am NOT talking about the class of people who stake 32 eth to validate nod…

I think it will become centralized, but not in the way that you think. We're likely to see base-layer transaction censorship, since nodes are no longer allowed to disagree with each other (or they'll be slashed), and far more than 33% of staked value is in the USA (and subject to OFAC compliance). I explained in more detail here: https://news.ycombinator.com/item?id=32532604

There's a third option you didn't mention in your comment: instead of signing the block with these banned transactions (option 1), or signing an alternate block with other transactions (option 2), the node could pretend to be offline for that block (option 3). Instead of being slashed, the node would then be subject only to the much smaller inactivity leak for these blocks.

The only question then is how indirect these sanctions can be. The node might not be allowed to validate a block with banned transactions, but what about the next block, which does not include these blocked transactions, only a pointer to a block with these banned transactions? What about the second next block, which has only a pointer to a block with a pointer to a block with these banned transactions? How many blocks until the link is tenuous enough that the validators under USA jurisdiction could validate again without fear?

Re: Mainnet Merge Announcement

#215

Post by dcolkitt »

api wrote: post #129

I have to say I'm very surprised, and I'm curious to see how this plays out. PoW must die, but I have a hard time seeing how this won't lock up a ton of currency and create its own plutocracy. Of course I suppose an argument could be made that a "least privilege plutocracy" with aligned interests is not as bad as many alternatives.

Why would this be any more plutocratic than any other capitalist economy? It's a mechanism where you can lock up ETH and get paid a steady return over time (about 3.9% today). Similarly the US dollar has treasury bonds where you can lock up USD and get paid a steady return over time (about 3.4% today).

Re: Mainnet Merge Announcement

#216

Post by tyrust »

spywaregorilla wrote: post #122
tsujp wrote: post #119

Earlier quoted context omitted.

there's just a small number of validators, which is just going to be the ethereum foundation and friends. There are over 415 000 validators. See: https://beaconcha.in/

Read the edit. Slashers. Not validators.

Any validator can run the slasher process.

Docs from one of the validator clients: https://docs.prylabs.network/docs/prysm-usage/slasher

Re: Mainnet Merge Announcement

#217

Post by AlexandrB »

howinteresting wrote: post #104
bowsamic wrote: post #34

Earlier quoted context omitted.

What if most of the miners collude to double spend tokens? The problem is the same. The difference is that proof of work automatically tends to centralisation due to its economy of scale: it's cheaper to add one more miner to your pool if you already have a big mining operation, but staking 1 eth is always staking 1 eth. Both PoS and PoW have the problem where you can buy out the majority if you have the power to do…

If most of the miners collude like that than ETH will lose most of its value. The game theoretic aspect of PoS is an important part of its overall security.

This is only true if ETH remains a niche security and not a widespread currency. A large population won't drop a commonly used currency because of ideological concerns like miner collusion. Indeed, they'll have massive incentive to keep using it since their saving may be denominated in it.

Re: Mainnet Merge Announcement

#218

Post by Geee »

jsheard wrote: post #123
spywaregorilla wrote: post #105

Earlier quoted context omitted.

So if I acquire 51% of validation power, and you catch me cheating, the person who gets to decide if I get punished is me?

Technically yes, but I think the idea is that it would be incredibly expensive and unproductive to pull off such an attack. You'd first have to acquire billions of dollars worth of Ethereum, and then launching that attack would torpedo any trust in the network and send the value of Ethereum to zero, lighting your billions of dollars on fire. The theory is that it's too expensive to realistically do for the lulz or ou…

It is not "incredibly expensive". Top 4 exchanges already hold > 50% of stake. Users give away their coins for free. For users, it's more expensive to run their own validator nodes than staking on a centralized service.

Re: Mainnet Merge Announcement

#219

Post by addaon »

immibis wrote: post #65
spywaregorilla wrote: post #56

Earlier quoted context omitted.

Who gets to run a slasher node and who gets to agree that its valid to slash a block?

I don't know the specific rules for Ethereum, but in general PoS systems, anyone can slash. All you have to do is submit two different headers for the same block signed by the same validator. This proves the validator is cheating. The person who submits the proof may get to keep some of the validator's slashed money, which is a good incentive to run slashing nodes as this can amount to thousands of dollars per slash,…

In a general PoS system like you describe, what prevents a bistable situation from emerging? If there is any significant level of cheating, then running a slasher is profitable, which makes cheating very unprofitable, driving it to zero. But in the absence of cheating, what's the motivation to develop, maintain, test, and operate a slasher node? Isn't this prone to long periods of time with negligible cheating where all the slashers die out, followed by one big malicious attack that isn't caught promptly?

Re: Mainnet Merge Announcement

#220

Post by chizhik-pyzhik »

kkielhofner wrote: post #211

What surprises me most about the Ethereum ecosystem is the software and release engineering of the implementations, especially the reference (now execution layer) go-ethereum. As has been noted elsewhere "the merge" has been in progress for years. Here we are approximately three weeks out and just this morning the (allegedly) working merge ready go-ethereum client was released. Given that the Bellatrix upgrade is sch…

> IMO the quality and process of the fundamental software enabling all of this is inexcusable - it's not like they're wanting for resources to do this right.

If anything, this just shows how complex the system is. Distributed systems are hard to write.

The recent Geth 1.10.22 bug only occurs on shutdown- an edge case I imagine they will be testing from now on.

Post reply on HN