Miners and Ethereans Clash Over ProgPoW
“We are in full support of the transition towards ETH 2.0, we’ve built a proof of stake pool, currently in the middle of test and will be release Q4 of this year. So we have no incentive to cause any split. If a split shall occur, it would be because of those who oppose ProgPow…
If ProgPow is not included with the Instabul [sic] fork, we will not cry over this. We will continue to support the health of the network the best we can. But we need for the West to understand this. Decentralization is extremely important to Ethereum, and if we allow for ASIC’s to take over the network, well you can expect nearly all mining to be concentrated here in the Mainland.”
The current eth-hash algo is designed to make asics not that much more efficient than GPUs, with no evidence there are any asics operating on ethereum.
There is in fact indirect evidence that there aren’t, or at least not of a significant quantity, as when there was a miners’ vote for ProgPoW, everyone was in favor.
Suggestions however are that they’re coming, with such suggestions made more than a year ago. Not without reason as there is an off-shoot of a prominent China based asics manufacturer, Linzhin, who has been trying to develop an efficient eth asic.
Just how efficient isn’t very clear as the asics is not out, thus there are no benchmarks, but unlike in bitcoin where they are 50x plus more efficient than GPUs, in eth they may be 4x.
Current big miners or mining pools seem to be keen on ProgPoW, so perhaps even 4x isn’t insignificant. It may be more, with current miners potentially risking going out of business if they do not upgrade their gear.
The plan is of course to make them go out of business through eth 2.0, but it looks like they might just pivot to running staking pools.
How that would work exactly remains to be seen as the pool incentives may be quite different with staking, but Proof of Work (PoW) is unlikely to be going anywhere anytime soon.
So naturally current miners would rather not have to buy asics, but one reason why bitcoin mining remains still fairly decentralized as shown by their inability to agree on the blocksize, is due to that continued hardware improvement which creates a race with miners rising and falling.
The absence of that natural competition, especially for a coin that is to Proof of Stake (PoS), may lead to current miners gaining significant staking influence as if you can’t compete with better gear, then due to economies of scale, there would presumably be continued concentration.
While in the asics market, there is fierce competition. Starting out is easy if you’re smart and have some upfront funds as you just buy the latest gear and are sort of in a level playing field.
With GPUs, you’d need purchasing power to get a good deal from AMD or Nvidia, with the student in his dormitory at a disadvantage.
While if he or she bought asics, they would still be at a disadvantage, but they would at least have a chance.
That student would be part of a vast global army that through the profit motive secures the network against any potential attack save for arguably by the government itself.
For ethereum, on the other hand, as GPUs are in such widespread usage, even a billionaire could maybe attack it.
It perhaps wouldn’t be pointless either because there isn’t quite a lot you can do in defense save for urgently linking eth1 to eth2 finality.
While if Linzhin, or perhaps the Chinese government or any government ordered asics farms to attack, you just change the algo and make their gear irrelevant.
There is little that can bring miners more in line than the suggestion of an algo change if it appears to have general support.
That’s for asics miners. For GPU miners, there’s little that can bring them in life full stop, except of course that generally they like making money and especially those with mining farm do rely on eth doing well.
So security, and the fact asics might actually be more decentralizing as bitcoin has seen more pools come and go in the same time as eth, is the main reason why many argue a coin shouldn’t just change its algo without a good reason.
In this case there are plenty of other reasons too, with ProgPoW apparently not good enough for even ETC.
That chain had a 51% attack which arguably was possible primarily because it sort of shares the same gear as ethereum: gpus.
In a way that’s kind of similar to merged mining, which has long been known to be insecure.
That’s because the dominant coin gets most of the gear, so to attack the smaller coin you usually need just 10% or less of the dominant coin’s hash.
In the case of ethereum the same applies but to a lesser extent because there are plenty of gpus already out there held by non-ethereans or even people who might not like eth much.
So the suggestion, and by miners, that the algo should just be changed willy nilly is a very curious one because while it might appear to be to their benefit in the short term, there are plenty of risks.
Just as lowering issuance to 2 eth instead of 1 might have seemed to be to their benefit, but it arguably wasn’t as eth’s price plummeted.
If there is a split over ProgPoW then it can get very messy as miners might actually find themselves without much support which isn’t really something anyone would want to see – well except bitcoiners maybe.
Moreover having an ETH and ETHP, or whatever they would call it, at the same time as eth 2.0 can actually get more than messy, so a split is probably unlikely as there isn’t general support for ProgPoW. So you’d think it wouldn’t be implemented.
Yet this keeps being brought up again and again on eth dev calls even though they previously said that wasn’t the right forum for discussing it.
So maybe miners are just trying to tire Hudson Jameson into saying yes, but he is obviously not the one that makes such decisions.
Such matters are instead ultimately decided by the free market if there is a chain-split, but ProgPoW is now probably far too late for Istanbul so this might be one of those things that just keeps being talked about arguably over far more important matters, like eth1x.