Ethereum Developers Spend an Hour on Deciding How to Decide


The latest dev call meeting of ethereum developers spent about an hour discussing how to make a decision on a potential ProgPoW proof of work algorithmic change.

“I am not comfortable to make a decision” – Piper Merriam, an eth dev said. Prior to that, Hudson Jameson, the chair of the meeting, asked: “do you think we need more data or we can just make [the decision]?” Greg Colvin answered: “we can just make it, it’s our job.”

Lane Rettig said a number of developers had stated they are not comfortable with making a decision. There’s a menu of options, Rettig said “one is decide to not decide, second is decide one way or another… third is implement both and maybe have some randomness that would institute a default and fourth is maybe some other course of action like auditing or bringing in some experts.”

Thankfully they didn’t go with flipping a coin. They didn’t decide to not decide either, nor did they decide either way despite some pressure to make a “f****** decision,” as Colvin put it.

Instead after some more discussion they went with miner signaling, or more correctly pool signaling with just two of them having close to 51% of the hashrate. That however was to be used as just one data point among many others as they went with the fourth option: auditing.

Hudson Jameson stated they had commissioned an audit on ProgPoW with a number of questions to be answered, including things like: “what changes in terms of hash power and miner balance in general?”

How they can answer that isn’t very clear because they’d need a survey of current ethereum miners, something that would probably take extensive efforts to do properly.

Hudson, however, said the audit might come back in March or April, so in a month or two. Meaning at best they’ll come back with only estimates.

The answer to some of the questions are already know. Like does it lean towards any particular manufacturer. The ProgPoW devs have already admitted Nvidia gear performs better due to a claimed compiler bug and because some computational requirements were harsher on AMD gpus. They’re now trying to address that balance.

Some of the devs on the call said ProgPoW wasn’t anywhere near ready as it requires quite a few things to be done before it can be considered for mainnet deployment.

Yet the eth1.0 meeting was solely used for this matter with a suggestion that this now may be revisited again once the audit comes back. So a decision was finally made to wait for the audit, which appears to be a very common sense approach but it took an hour for it as some wanted to decide now what they will decide once the audit returns.

This matter was apparently not on the agenda at all to begin with, but then some flooded the ethereum subreddit with thread after thread about ProgPoW so they got what they wanted.

Some even begged Vitalik to save the poor miners in their massive farms even though Vitalik had previously raised the question of whether ethereum is not more secure with asics, stating last month that if miners misbehave they can just brick the asics, but “If attacker using gpu, there’s nothing.”

This question of whether the network should interfere with the algorithm at all and what might be the consequences of such interference as far as security is concerned was not a matter of discussion in a yet another ProgPoW devs call.

Nor was there any discussion of FPGAs by Xillinx and Intel with FPGAs being programmable asics. Most importantly, there was no discussion about eth 1x even though there are tons of decision to be made there, decisions that actually matter for usability, scalability, dapps and the entire ecosystem rather than for just some miners vying with each other for a bigger piece of the pie.

Miners who don’t fund one developer as far as we know, or any ethereum project, or in anyway contribute to the ecosystem besides the burning of energy for profit.

Alexey Akhunov, who is working on rent storage and other improvement proposals for Ethereum 1.x, said in the previous dev call last month that he was not aligned with asics or no-asics, but from his point of view concerning his task of developing and designing eth 1.x, the more hardforks there are planned, the harder for eth1x.

In that call developers decided that dev calls were not the proper forum to make a decision on ProgPoW. Yet effectively all of yesterday’s dev call was spent on ProgPoW.

That’s presumably because some have a significant interest in getting this algorithmic change into the mainnet. Some, including AMD miners and Asics miners but more importantly some devs working on actual improvements have an interest in resources being focused elsewhere.

Yet, whichever side, Colvin perhaps is right. Even a bad decision might be better than no decision, but the problem is that this concerns a fundamental aspect of ethereum’s security which can have significant implications if the hashrate drops considerably, so opening ethereum to a potential 51% attack.

That may seem far fetched, but there have been a number of 51% attacks last year, including of ETC which is quite similar to ethereum. It is unlikely that was a testrun for a potential eth attack, but that is one unknown unknown among many others when considering this question of an algorithmic change.

If the ethereum ecosystem wants to take that risk then that’s their choice. There should be a well publicized coinvote lasting for a month with at least 5%-10% of eth voting in what looks like a non gamed manner. If there is a clear majority one way or another, say 75%+, then that’s the decision.

If there is no clear majority then ethereum would be in unchartered territory for there has been no main-chain upgrade where there was even a 40-60 split, let alone in the 50s.

So that’s how you make a decision that can be considered fair and can be considered objective although it would be open to some criticisms, but there isn’t really any better decision making mechanism when it concerns non-technical questions that are controversial.

It is a better mechanism than some council in any event which spends an hour to decide they’re going to wait for an audit as if the newest invention is that of having full information before contemplating a decision.

On a final note, there are numerous teams within ethereum which are sort of compartmentalized. The eth 1 team, for example, kind of has almost nothing to do with the eth 2.0 team. So this stupendously laid back approach does not necessarily reflect on the latter, but we’ll see whether that’s actually the case once they start rolling out the PoS Beacon which should begin next month.



Leave a Reply

Notify of