Consensus Mechanism


#1

Anything related to æternity’s blockchain consensus mechanism.


#2

Hi, Vlad
I am Peterson who are operating Korean Community.

I already know that a master node doesn’t exist on AE. And the Staking is for Oracle betting and Governance.

When I saw the white paper “Consensus mechanism” section in the 5 page, I read the following sentence.

“Therefore Æternity is going to use a novel hybrid Proof-of-Stake Proof-of-Work algorithm, leveraging the benefits of both. . Independently from this, PoW is going to be used to issue new aeon tokens.”

Looking this sentence, it looks like it is able to get aeon by using POW. But what is the mean of “a novel hybrid Proof-of-Stake Proof-of-Work algorithm”.

In other words, what is the Proof-of-Stake in this sentence. Is it for Oracle?? Or something else?


#3

Hey! Thank you for taking care of the Korean community! :slight_smile:

Regarding your question -> the novelty comes from the fact that the PoS will be used only for governance. The PoS will be used primarily in the form of users voting with their token balances. There will be no staking, so I see how PoS could be confusing. At a later stage the PoS could also be applied to oracles.


#4

This makes sense. I think most people see “PoS” and think they can earn AE as rewards very easily by simply keeping their wallet open and running on a laptop and staking like most other PoS coins out there right now.

So when they hear that there is no staking they get confused, and think that they will not be able to earn reward AE passively, and that mining would be too technically or mechanically out of their reach. That “mining” would be too complicated.

But from what I understand, “mining” with AE will be much more accessible to the average user with a decent laptop, correct? That with the Erlang Cuckoo Cycle algo, mining will be more RAM dependent, and that one will be able to easily mine AE via it’s wallet using even a laptop’s resources, and receive reward AE?

Or will they have to run a full node on a Linux machine or vps or something to do this??


#5

Using »Proof of Stake« to describe the mechanism is very confusing indeed.

@AELifer, you are correct with your assessment that the cuckoo cycle PoW does not lend itself as much to specialised hardware, i.e. ASICs, but a powerful GPU will still have a considerable edge over a CPU, so while you can mine with a laptop and not have to compete with ASICs, you will still have to compete with GPUs.

Regarding your last question: unfortunately there are no Windows builds for Epoch currently.


#6

Thank you for reply, Vlad.


#7

How does your statement relate to this one?

A hybrid ASIC solution for Cuckoo Cycle pairs a bunch of DRAM chips with a small low-power ASIC, which needs to run just efficient enough to saturate the limited DRAM bandwidth. In terms of solutions per Joule of energy, this might be reasonably efficient mining platform.

(from https://github.com/tromp/cuckoo#hybrid-asics)


#8

Specialised hardware will always outperform general hardware, at which point it comes down to the margins you can get with the specialised hardware. The claim is, that for cuckoo pow, this margin is not big enough to make the development and manufacturing worth it.

However, I’m not aware of any studies investigating just how big or small the margin actually would be.

Edit: Oh and earlier today, an article was published by the lead developer of Sia, which gives a great overview, as to why the whole endeavour of »ASIC resistance« might be futile anyway: https://blog.sia.tech/the-state-of-cryptocurrency-mining-538004a37f9b


#9

What’s happened to the Cuckcoo Cycle, is that no longer being used by Aeternity?


#10

We’re still using cuckoo cycle pow. Nothing has changed there.


#11

Awesome, can you briefing explain the link between the cuckoo cycle & bitcoin-ng? They’re being used together?


#12

Thanks for the info and the link, very interesting.

The problem with the margin argument is that it only holds temporarily. If the market cap of the ecosystem increases sufficiently enough or yet more efficient ASICs are developed, it will break down. I.e. while it might be possible to devise a POW scheme which makes ASICs not worthwhile now, I don’t see how one can plausibly promise this for future developments.


#13

When using Bitcoin-ng, miners need to solve a computational puzzle in order to produce blocks, just like in the original Bitcoin protocol. The computational puzzle (PoW) for Bitcoin is based on finding special SHA256 hashes. The original proposal for Bitcoin-ng used the same puzzle as Bitcoin. We opted to use cuckoo cycle instead of SHA256.

So Bitcoin-ng is a consensus protocol/algorithm, where cuckoo cycle is used to solve some of the problems, that need to be addressed to make the system work, e.g. leader election.

Does that make sense?


#14

I wasn’t really thinking of profitability, although that is sort of implied by the computational margin as well.

I definitely share your assessment that guaranteeing permanent ASIC resistance—i.e. ASICs never having more than a e.g. 2x advantage over a GPU—is most likely not feasible. You can commit to playing a cat and mouse game, in the hope that you end up ahead of the hardware.

It’s a complex topic and not just a binary matter of ASICs are bad and resistance is good, where a lot of arguments are very emotional and not always rational.


#15

Yes! Great explanation, thanks very much, I had misunderstood the role of the cuckoo cycle.


#16

Power concentration is bad for systems which are supposed to be decentralised, and profitability of ASICs tends to generate such concentration, as we can see on the Bitcoin chain. So trying to find ways out of that is worthwhile.


#17

Finally had time to properly read that article. Pretty depressing really. If PoW by itself cannot ensure decentralisation because of the stated economics of ASIC manufacturing, and other “structures and schemes” are required to effect decentralisation, then what is the use of PoW in the first place?

I mean, the whole purpose of PoW is to make it too expensive for any one party (or coalition) to have enough control to manipulate the chain single-handedly. This is not true anymore if few or even just one party controls almost all of the hash rate.

In the article they argue that such a party cannot afford to play dirty because that would lead to hard forks and people going elsewhere. However, that’s true for any centralised system (which it then effectively is): if the central party becomes untrustworthy, people move elsewhere. But this is how all traditional and centralised systems (in an open market) have worked.


#18

Interesting discussion. I just wanted to add that staking will be available for answering oracles, which is of course not a consensus mechanism.


#19

Just so you know everyone - ssh is Sascha Hanse - æternity’s lead blockchain engineer. :slight_smile:


#20

Yes good to know, thanks :slight_smile: