Hacker News Reader

Comment by msoad | Hacker News Reader

Isn't PoS inherently against decentralized nature of cryptocurrencies? You have to have money to use money

PoW is one level of abstraction away from "you have to have money to use money"; "you have to have money, to buy compute, to use money". Additionally, compute resources are subject to centralization.

There is a crucial difference between decentralized and equitable. A decentralized blockchain needs some mechanism to guarantee perpetual operation in a trustless environment, like requiring expensive computations (proof of work) or showing you have "skin in the game" (proof of stake) in order to add transactions to the shared ledger. If you don't have such a mechanism, there's no way to tell valid from invalid transactions, and the blockchain falls apart at the seams.

What it sounds like you are suggesting is that cryptocurrencies should be equitable: everyone should have the same rights to own part of the blockchain. This isn't necessary for a decentralized, trustless, distributed transaction ledger, nor is it likely ideal. What would be the value or utility of a ledger which absolutely anyone could add to at anytime for any reason? You would need to solve 2 problems: 1) distributing the tokens fairly and evenly among every person living and yet to live, and 2) distributing henceforth unfathomably performant supercomputers to all of these token-holders-jus-sanguinis. Any analysis of either of these would very quickly approach insurmountable problems.

There are fundamental constraints to computers and computer networks that have to be addressed in order for these technologies to exist. They preclude any sort of utopian currency of heaven where everyone wins, just by the nature of our physical universe and today's understanding of computation.

With regard to your first problem, it may be hard but should result in a more thoroughly decentralized system, no? And on the second, I thought PoS is orders of magnitude less compute-intensive?

Not any more than PoW, which has worse economies of scale in which the richest get richer at a faster rate as they can lock up supply of available hardware.

The big difference is that hardware depreciates. Proof of stake requires none other than a single purchase to be gifted the right to perpetual seignorage. Staking ETH once versus constant operation of a mining farm is a non-trivial delta in effort.

…and to maintain (or pay a host to maintain) a node.

I suppose so. If you're payment to an agent to carry out either task, the downstream complexity of the task is irrelevant because the difficulty to you is the same.

It costs less to stake, but it pays less too. Ethereum's staking rewards are an order of magnitude lower than its mining rewards.

The minimum amount required to stake is 32 ETH. That’s $75k at today’s price. There are cheaper was to “stake” but those are not really “ETH PoS” staking. So no, it does not “cost less to stake”.

What are you comparing $75k to? My understanding is meaningful mining under PoW requires buying expensive mining hardware that goes out of date quickly. Under PoS there is no such depreciation...a laptop from 2020 will likely be able to run PoS in 2040.

I personally have been mining ETH for years with equipment I bought for $20k in 2017 - 2018. I’ve made many times my investment back. That same equipment is still running profitably today.

Are you ASIC or GPU mining?

I have 30 GTX 1070 GPUs.

And what do you plan to do after the switch to proof of stake?

If I have enough, I will run a validator. If not I will probably do some defi like lido + curve + yearn. I may also mine something else like flux or raven.

There are ways to have a fractionalized stake, essentially the PoS equivalent of participating in a PoW mining pool. They call them staking pools, which seems logical enough.

https://stakingether.com/

https://rocketpool.net/

https://beaconcha.in/stakingServices

I agree with you of course, but those are not “ETH PoS” staking, strictly speaking. They are just layers on top. You definitely make more running your own validators.

There's initial capital expense, and ongoing expense. You're of course correct that the minimum capital is lower with mining. However, given a $75K initial investment, a miner has much higher ongoing expense.

That's not necessarily true. Some people mine with 100% solar power and their ongoing expenses are almost zero.

So, my original point was to respond to the above comment: "Proof of stake requires none other than a single purchase to be gifted the right to perpetual seignorage."

My claim was that the lower ongoing expense of staking is compensated by lower rewards, so that both systems have similar profit for similar capital outlay. In setups where mining expense is almost zero, the higher rewards of mining give it much more seignorage than proof of stake.

There is one thing: decentralization, and there is another: concentration of ETH ownership.

Concentration of ownership means a few entities own most of the ETH. Up to a point (where one entity owns all the ETH, and therfore ETHUSD goes to zero), who cares, as long as there are still enough token in circulation to conduct business?

Decentralization OTOH means that no one can prevent me from executing a transaction on the chain.

With POW, these two things are basically unrelated: concentration can be arbitrarily high, I can still transact.

However, IIUC, with the aptly named POS, concentration of ownership brings concentration of voting power and therefore large holders become theoretically capable of censoring transactions ... not great.

And I don't believe there is any mechanism in place to prevent ownership - and therefore voting power - to reach arbitrarily large levels.

There are over 200,000 ETH PoS validators running already. I think that’s pretty decentralized.

How many banks are there?

In the US, probably around 10,000. US Banks all operate by virtue of a banking license issued at the state or federal level. Banks must do as they are told by these centralized authorities or else they lose their license.

https://www.statista.com/statistics/184536/number-of-fdic-in...

4,300 or so FDIC insured banks.

But um I'm not sure it's so different from Ethereum, where a mining node must implement the protocol or else be unable to participate.

Decentralization is about not needing anyone’s permission to participate and being free from censorship. You don’t need good credit or even ID to open an account and no one can seize your funds because of debts, fines, or taxes. Of course validators have their slashing rules, but those are programmatic. Not some judge in divorce court, for example.

> Decentralization is about not needing anyone’s permission to participate

But you do need permission, which is implicitly conferred by following the protocol. It's really quite similar to traditional banks, it's just that the protocol is digital/mathematical rather than legal.

Indeed. In my experience the legal system is highly biased and corrupt.

I'm wondering if we should care as long as there's reasonable assurance that they won't cheat when executing transactions? You don't need full democracy where everyone gets a say, just enough decentralization so they're unlikely to collude to do whatever bad things you're afraid of.

Reply on news.ycombinator.com