Topic: Ethereum Anti-ASIC fork, is it the right time for bitcoin too? - page 9. (Read 2439 times)

For an algorithm to be "cracked", we'd first have to agree that designing hardware to perform a task more efficiently is effectively breaking some sort of implied lock.  I'm still yet to be convinced there's a lock to break, let alone that ASICs are somehow breaking them.  It's unlikely we're ever going to see eye-to-eye on this.

Perhaps if an algorithm was specifically designed with ASIC resistance in mind, then you could make that claim.  But that doesn't apply here.

How is that? Taking over of hashing power is just a matter of decision making for a company like Bitmain with ASIC mining being dominant while there is no way for any entity in the globe to dream of controlling like 10% of gpu power. GPUs are globally distributed and controlled by hundreds of millions of people and manufacturers like AMD or NVIDIA have no meaningful presence in cryptocurrency marketplace and there is almost no possibility for this scene to change regardless of what happens next.

Let's don't argue about indisputable facts: GPU is afar better choice than ASIC for any cryptocurrency.
Actually ASIC is a crack against cryptography, it has always been since WWII and nothing has changed, when a cryptographic algorithm get ASICed, it should be considered a failure and fixed instead of being justified as 'inevitable', 'not a big deal' or even 'a good thing'!
It is just ridiculous how is it possible to have a cryptographic system of any kind being cracked by a specialized circuit and considered safe meanwhile?

ECDSA is definitely the more worrisome factor, but I'm indeed referring to using quantum computers for mining.

While quantum computers don't have any apparent computational advantage over classical CPUs as far as SHA256 is concerned, they are also not especially disadvantaged. With classical computing reaching its physical limits and quantum computing just getting started, it seems rather likely to me that quantum computing (assuming it doesn't hit a wall itself) will outperform classical CPUs / GPUs in terms of SHA256 hashes within our lifetime -- with ASICs offering a much wider margin of safety.

I have to admit though that this point becomes utterly moot under the assumption that Bitcoin's future PoW scheme is as resistant against QC as it is against ASICs.



I'm aware of your proposal and didn't mean to distract I just think it's worth keeping in mind that not ASICs per se are the problem, but the centralization that they enable (ie. nothing is gained if we get rid of ASICs only to be re-centralized soon after).



Is it though? The GPU industry is essentially a duopoly of NVIDIA and AMD, with Bitcoin's GPU mining era having been an AMD monoculture. Additionally the 2017 bubble had enough of an impact on the GPU market that regular gamers were facing a supply shortage and prices inflated way beyond the original release date prices. And that was just alt coin mining, even excluding Scrypt-based coins.

GPUs are definitely wider distributed than ASICs, but don't underestimate the market impact of crypto. Given a crazy enough market high-end GPUs (ie. the ones that are relevant for mining) can become just as hard to attain for the common mortal as ASIC miners.



Funny, I personally assume quite the opposite as I'm instantly reminded of the BCH / BTC hashrate fluctuations and the recent 51% attacks on minor alt coins. Not necessarily a problem for Bitcoin, but definitely not stabilizing for the ecosystem as a whole.

That's just my thoughts though, what leads you to the conclusion that Bitcoin sharing hashrate with alts would be stabilizing?



If your investment is at stake and you can't use it for anything else (as is the case with ASICs), loyalty becomes the rational choice. Granted, it's the kind of "loyalty" that only lasts until the hardware becomes obsolete, but that's still a couple of months of hashing that you would otherwise need to continuously compete for.



Requiring more confirmations is the obvious solution, but also an incredibly impractical one. Having your customers wait for a day worth of confirmations on the off-chance that a someone is about to mount a 51% attack is rather crippling, even if you only apply it to larger transactions.

And WE have to disappoint you, sorry.

Giving up with rational behavior assumption gives us nothing other than chaos. Without such an assumption we have to forget about game theory and without game theory, we are left alone with almost nothing to solve Byzantine Generals Problem in a trustless, open environment.

I appreciate your courage for thinking out of the box, but I think it is not the right place to discuss such an elementary issue. If you are really interested in developing a new class of cryptocurrency which is not based on rational behavior of participants, you are welcome to do some paperwork including a mathematical model comparable with what bitcoin is built around. My advice: Don't even try, you fail.

I do not agree with people saying ASICs are a threat for x coin, it's not true. the actual treat is the control of hashing power regardless of where it comes from.

if we fork against ASICs those with access to enough GPUs will still be a  treat to the blockchain , if we fork against GPUs, same thing applies to CPU, if we fork away from CPU those with most pens and papers + brain will be the new treat.

pos does not offer a much different protection against "power control"

and I need you guys change this point of view [1] - especially when you try to disable ASICs in a crypto network. in the field of decision making, scientist follow new patterns that match better to human mind that first came with Behavioral Economics:

"Behavioral economics studies the effects of psychological, cognitive, emotional, cultural and social factors on the economic decisions of individuals and institutions and how those decisions vary from those implied by classical theory"  [2]

"conventional economics assumes that people are highly-rational—super-rational—and unemotional. They can calculate like a computer and have no self-control problems."  "People often make poor choices—and look back at them with bafflement!" "We do this because as human beings, we all are susceptible to a wide array of routine biases that can lead to an equally wide array of embarrassing blunders in education, personal finance, health care, mortgages and credit cards, happiness, and even the planet itself." [3]

if cryptocurrency is going to be a competitor of monetary, banking or payment system, needs to include a wide range of irrational behaviors (misbehaving [4]) in the equation.

[1] https://bitcointalksearch.org/topic/m.48113411
[2] https://en.wikipedia.org/wiki/Behavioral_economics
[3] https://en.wikipedia.org/wiki/Richard_Thaler (He won the Nobel Prize for economics in 2017)
[4] https://en.wikipedia.org/wiki/Misbehaving_(book)

In cryptocurrency we have no assumption about loyalty, all in all it is about being rational instead of loyal. I know you are aware of this point I just don't get it why should you make such a weird argument tho.

@HeRetik has already mentioned the interesting consequences of mining industry being homogenous mainly in terms of using same hardware, gpu. My first impressions are encouraging. I think we will have a more stabilized mining scene for both bitcoin and alt coins.

Let's forget about 50%+1 attack, not everything is about such an attack and it will never happen for a big fish like bitcoin or ethereum and smaller ones should take care of their business by smart measures, I've proposed one earlier: require more confirmations for more precious txns.

There are a lot of other socioeconomic problems and issues to be addressed:

-Is it possible to have current ASIC miners' support?

-Is it possible to make a hard fork without leaving behind a significant portion of users and miners?

-Do we have to get rid of pools simultaneously?

I don't follow how ASIC or gpu based PoW would have anything to do with QC vulnerability. Actually I believe they don't.
I assume you are speaking of SHA256 immunity to QC attack but it is the same for ProgPoW and many other gpu/cpu friendly algorithms, they are as resistant to QC as sha2 if not even more. QC is supposed to crack ECDSA encryption in the transient phase in which pub key is disclosed by user (to claim the utxo) and has nothing to do with mining.

You are absolutely right, and you may be already aware of my other proposal, Proof of Contributive Work, PoCW which is presented in a topic titled as Getting rid of pools ...
Obviously, we are discussing a hard fork here and it is totally possible to have both at the same time but we need to discuss each issue separately. I'm afraid getting into pooling problem would be a distraction here.

GPU industry is far more competitive and healthy than its ASIC counterpart and more importantly, gpus are distributed around the globe more evenly. A gpu friendly algorithm of any type has a large memory footprint (to resist being cracked by ASICs like what happened to bitcoin) and memory banks are a rare resource with limited global supply, a scene that is unlikely to change for near future.
IMO, there is nothing to worry about gpus being monopolized, nobody could afford such a huge investment because the demand for mining is a fraction of the total demand for gpus and manufacturers won't sacrifice their market for a tricky one time business.

As of altcoin gpu miners, I believe it is really interesting because they can act as another stabilizing factor.

I totally get your point, but I also could remember how two things happened after second bailout for banking system in 2009. 1st, the bitcoin. 2nd, a phrase about climate change with this message: "we can't eat money" [1]. both targeting the need of tighter regulation in banking and climate areas. if we all just think about money, there will be a day that we all lose hope and trust in a moment. bitcoin is important because it is going to be an answer for the broken trust of people about their future, now the same problem (everybody just look for money) is going to break it down.

and also look at these calculations and the inversion of incentives for miners of ETC:

block interval= 10 second
block reward= 4 ETC (20$)
blocks (last 24h)= 6,108

block reward (last 24h) = 24,432+24.45+34.13+34.13 ETC ($105,506.96 USD) [2]
stolen amount (less than 24h) = $1.1 million worth of the currency [3]

and put them all behind the tweet of ETC team:

"We are working with Slow Mist and many others in the crypto community. We recommend exchanges and pools significantly increase confirmation times (400-4000+)"

you see? if I be a dishonest pool and have 51% of the process, then I could gain 10 times more benefit that being an honest pool at the same time. I call it the inversion of incentives. who and how could guaranty such an inversion of incentives never happens to bitcoin?


being mathematically broken is not the only matter of security, being available is also important in the context of security. ASIC devices sometimes could bring a kind of denial-of-service to the network (51% attack). nobody is happy with this situation, but this is a bitter fact that we should face it.


misunderstood. I agree that equipping by ProgPoW would not be that hard, but getting the desired result is another issue here. you know, when we have ASICs for sha256 and some other ASICs for ethash, none of their owners could perform an attack against the other group. but, if we all destroy our ASICs and get back to GPUs then we have a large army of multi-purpose processors that have no loyalty and just look for a coin that bring more profit for them within an specific period of time. then imagine a guy (proof-of-keys [4]) that suddenly appears in youtube and asks all miners to join his pool that is planning to attack bitcoin, just for fun or any reasons else.. this is where we need the REQUIREMENTS and CONSIDERATIONS part..


[1] https://me.me/i/when-all-the-trees-are-cut-down-when-a-the-3803512
[2] https://bitinfocharts.com/ethereum%20classic/
[3] https://www.technologyreview.com/the-download/612728/hackers-just-stole-1-million-from-the-ethereum-classic-blockchain-in-a-rare-51/
[4] https://www.proofofkeys.com/

I personally am still pro-ASIC mining for various reasons. Mainly because it increases the stakes that miners have in Bitcoin; to a lesser extend because it decreases the vulnerability of Bitcoin's PoW to future quantum computers (under the premise that early quantum computers will be even more centralized than early ASIC miners).

Given the size of the industry nowadays I also have my doubts that a mere switch to GPU / CPU mining would help decentralization all that much; at least not without additional measures to make pool mining significantly less attractive. Regardless of this, economics of scale would still apply and whether big miners buy ASICs or GPUs wholesale seems to make little difference to me. However I'm not quite up-to-date with the current mining hardware market and Bitmain's stranglehold may be worse than I think.

It would be interesting to see what sharing mining hardware with Bitcoin would mean for alts though (and vice versa).

I don't think ASIC supporters are the same as btc holders but I do agree with your point, we need to do something about the investments made in ASICs because it is what cryptocurrency is about: responsibility and consensus but we need not to let in greed and centralization instincts, as well.
So, this is what I propose:

Firstly, I think we should draw a line between ASIC manufacturers and ASIC miners (all btc miners) to distinguish them, both have invested in ASICs but in very different senses. I feel less responsibility for Bitmain as this company has made enough up to now but miners are another story. I think there is a solution for keeping them safe and profitable enough to rebuild their business in the correct direction: gpu mining.

I've been discussing it in different situations, we don't need to force ASICs out immediately, with a very little effort it is totally feasible to have both SHA256 and ProgPoW active for a very long time.

Unlike what many people think btc miners are not happy with ASICs, difficulty surges with a crazy rate even when price is dropping bad! ASIC manufacturers install proprietary mining farms and compete with miners (their customers!) with unfair advantage of using newer technology earlier than it has been even announced to exist. In 2018 bitcoin price tilted down from 20K to 4K and meanwhile network difficulty has surged like 2.5 times! It nets to a 1200% loss of profit for miners! No, miners are not happy with ASICs they are running, they are slaves of Bitmain and they need to be set free of this slavery.

For this to happen, I've proposed a multi-algorithm PoW using both SHA256 and ProgPoW with a 10:1 ratio respectively in the start and a predefined re-adjustment of this ratio in a 2-3 years period to have it at 2:10 i.e. allocating 1/11 of blocks to ProgPoW in the beginning and improving it up to 80% within 2-3 years, by applying just one hard fork.

This will be even more profitable for ASIC miners because puts an end to Bitmain (and others) projects and decreases supply of ASICs substantially. It guarantees a more stable and continuous profitability after the first shock . Miners would be able to retire their ASICs gradually and instead of buying new ASICs to invest on gpus.

Although I'm personally running a small mining farm, in spite of some positive feelings about such a transition from sole economic perspectives, I've not done a thorough assessment on socioeconomic impacts involved, I confess, and I can't speak on behalf of mining community.
It will be highly appreciated if more serious mining stakeholders would tell us about how such a proposal looks to get support from their side.

Of course there is such an allergy, but that part of my reply is not against your decent argument, I was just trying to express my objection to people who believe that there is no anti-ASIC algorithm, people like the ProgPoW inventor.

My point is a cpu-friendly algorithm is feasible to implement but a network totally based on such an algorithm is not necessarily secure and decentralized as taking control of large processing powers by botnets is feasible as well. Monero case, ase you've correctly reminded, proves it.

On the other hand gpus are interesting devices, taking control of enough gpus to cause serious damages is not feasible either by implanting botnets or installing large mining farms.

I have a use case for cpu-friendly PoW tho: attaching work to transactions when they are generated by wallets which is another topic and for now we could simply ignore it.

I've to disagree with your argument about it being such a HUGE project form the technical point of view:
ProgPoW is not that immature, bitcoin core code is not that messy and Ethereum will provide a test bed. We need to get ready.

I think there are a few socio-political issues to be addressed tho. @ETFbitcoin and @mu_enrico have reminded some.

As of ProgPow being gpu friendly, the inventor has argued that it is somehow inevitable because she believes there is always a possibility to crack any algorithm by ASICs and it is more safe to design it for an ASIC from the beginning and gpus are ASIC! They have a lot of specialized circuits and PogPoW utilizes them to make it almost pointless to design a chip just a bit more specialized than a conventional gpu, smart girl!

I don't believe in the first part of the above argument, that any algorithm is breakable by ASIC, you can always have a truly memory hard algorithm that resists parallelization, actually I've been working on one for a while, yet I do believe that gpu is a better choice than cpu because of another reason: sequential, cpu-friendly PoW is vulnerable to botnets as hackers would be able to get control of large processing powers and damage profitability of miners, for the least. A cpu-friendly algorithm is more adequate for hashcash like applications where wallets add work to transactions (and not blocks as the case with miners) and contribute to security partly with a replace-fee-by-work policy.


I'll come to your other points later

I think we cannot do the anti-ASIC fork in bitcoin since companies already invested a lot of money in SHA256 ASIC. If you do the voting (1 BTC = 1 ticket), there is a strong probability that ASIC supporter would win. Moreover, a fork without miner consensus will lead to more negativity towards BTC. For example, BSV/BCH might say miners don't matter in BTC.

I support algorithm hard-fork if and only if SHA256 is no longer secure.

if we follow the recent 51% attack (5-Jan-2019) on Ethereaum Classic (ETC) [1] we could say now this is even late for such upgrade. ASIC devices were always harmful to the PoW, and like many others, I have explained it in several threads too [2]. there is also detailed information about ProgPoW in link bellow:

https://medium.com/@ifdefelse/understanding-progpow-performance-and-tuning-d72713898db3

but this would not be enough to upgrade into a ANTI-ASIC algorithm. for example, using the RSA alone as public key infrastructure is not enough to provide security. there are lots of REQUIREMENTS and CONSIDERATIONS [3] that we should take during implementation to provide the security.. generally decentralized projects need to design new data structures and work flows that are more focused on opportunity cost factor of providing 51% attacks. I could see how users of ETC are in trouble these days.

[1] https://medium.com/@TokenHash/why-i-am-still-optimistic-about-ethereum-classic-etc-ef1ebe13c44b
[2] https://bitcointalksearch.org/topic/m.48111338
[3] https://tools.ietf.org/html/rfc3447

UPDATE: hope someday soon we see that there is a PRIORITY TAG added to the tasks that bitcoin should take ahead and people in community vote on them. this would be a good idea for better decision making.

There's no wrong time, but it might be best time, especially following momentum sometimes is good idea. Even so, there are many problem that must be overcome such as :
1. Get community support, especially most people are against hard-fork
2. Make sure Bitcoin network isn't vulnerable during transition, there are few things that should be thought seriously such as initial difficulty
3. Get miner support, which is unlikely because SHA-256 ASIC pretty much only can mine Bitcoin. Even BCH only have 3% of BTC's hashrate, so their ASIC essentially become expensive heater.

But while ProgPoW is great idea, i don't like the fact it's utilized for GPU.

In governance Ethereum is far more centralized than bitcoin, they have Vitalik both as a celebrity and a spiritual leader and believe it or not they have a roadmap  

IMO, a cryptocurrency with a leader,  is not reliable in the first place, but when the leader turns out to be a PoS believer in charge of a PoW coin things get even more confusing. I believe that Eth 2018 falling down 3 times worse than bitcoin has some thing to do with this fact.

Still there are good news as well: Vitalik is growing up and stepping down, well, not officially and completely but there exist signs. Most importantly, in January 5 latest Ethereum core dev meeting ended with a long-waiting admission, tentatively tho, of implementing ProgPoW as an anti-asic algorithm to retire Ethash. ProgPoW is designed to utilize gpu strengths such that it is almost impossible for asic manufacturers to build a considerably more efficient chip for mining it and not ending to to a gpu design project.

It is an important event in cryptocurrency and I think we will be witnessing a new wave of debates and discussions in bitcoin community regarding the situation with ASICs and the potentials for an anti-ASIC fork.