Partnered With:

Why is PoW the only choice for the value storage public chain? Hear what core members of Nervos and the Bitcoin community have to say

Read this article in 29 Minutes

PoS vs PoW

Original title: "PoW v.s PoS: Why PoW is the only choice for value storage public chains? Hear what core members of Nervos and the Bitcoin community have to say! "

Original Source: Nervos Chinese Community

On November 5th, Ethereum’s ETH 2.0 mortgage contract was officially deployed, and community mortgages were opened 32 ETH became a node, and it is expected to raise at least 524,288 ETH and 16,384 verifiers before December 1st. At the same time, Vitalik also wrote an article to speak out for PoS, telling everyone why ETH 2.0 chooses PoS. Up to now, the funds raised by ETH are about 294,432 ETH, with a total of about 9,200 pledgers. (Rhythmic Note: The Ethereum 2.0 contract address has completed the minimum pledge quantity requirement on November 24, and the genesis of 2.0 will start on December 1 as planned.)

For a long time, Nervos has always adhered to asset security, and also from Bitcoin and The success of ETH 1.0 sees the security brought by PoW and the fairness of entry. And for a long time, members of Nervos have expressed why we support PoW on various occasions. Recently, the core members of Nervos also expressed their views on PoW:

Jan: Punishment mechanism cannot save PoS

Jan believes that the punishment mechanism of PoS and the different cost of obtaining ETH will cause many innocent asset holders to suffer, while PoW will not cause such problems. On September 13, Vitalik made some remarks about PoW on Twitter:

In a PoW system, a successful attacker can continuously target When the system is attacked, the system cannot delete the hardware device used by the attacker. From the perspective of the bottom layer, this is where PoS is stronger than PoW.

Nervos Chief Architect Jan raised his doubts:

(Vitalik's) argument is wrong

1. For systems built on economic incentives (PoS and PoW), the person who controls 51% of the computing power/tokens Better to cooperate than disrupt. If we disregard these assumptions, then the "slashing" penalty mechanism cannot save PoS, because the attacker can short-sell the tokens in the external market to cover the slashed tokens, and then repeat this process continuously.

2. In fact, the attacker does not need to have 51% of the total tokens, it is enough to control 51% of the validator nodes. And the destroyed deposit may belong to someone else (note: such as an exchange), and the punishment mechanism makes things worse at this time, because the person whose tokens are destroyed is innocent.

3. DeFi on PoS will exacerbate this situation as discounting of PoS deposits will further reduce the cost of attack.

4. In the PoW 51% attack, changing the hash function is not the only solution. A whitelist of temporary miners can be formed through community consensus. (Vitalik's solution to most PoS problems), and use it for nodes until more "honest" computing power joins the network. In this case, PoW can do it at least as well as PoS.

At the beginning of November, Vitalik published "Why PoS is awesome" (https:// www.chainnews.com/articles/801734528092.htm) , Jan also commented again:

This is totally wrong.

Closing and weak subjectivity are called unimportant issues, but instead emphasize efficiency, it is better to use a centralized system directly, the efficiency is the highest . Calling a problem that is generally accepted as unimportant is pure deceit.

Zhang Ren: The security of PoS depends on many uncertain security assumptions

Then, Zhang Ren, a Nervos researcher who once worked in the Blockstream Lab, which knows Bitcoin and PoW mechanism best in the world, (everyone calls him Mr. Ren), how do you view the PoS issue? Here we briefly sort out Zhang Ren's arguments for everyone:

More security assumptions + great security gap between ideal and reality = great system vulnerability!

Zhang Ren once published a topic at the San Francisco Blockchain Center last year: In the speech "What PoS Cannot Achieve (But PoW Can) And What PoW Cannot Achieve, Either", he once proposed:

System vulnerability = (ideal security attribute - actual security attribute) + security assumption

From this formula we can see that when we want When building a better blockchain, our key goal should be to reduce the vulnerability of the system as a whole. To achieve this goal, we should achieve "compared with the previous design, the newly added security assumptions are less than the newly added security attributes"

The operation of the proof-of-stake mechanism requires a scarce resource as the basis. However, unlike the proof-of-work mechanism, if there is no trusted subject, the network using the proof-of-stake mechanism cannot tolerate high latency. This means that in some cases, proof-of-stake systems face several attacks that can make it impossible for new participants to discern the real network history.

In order to solve this problem, a system using a proof-of-stake mechanism must use a system signed by some trusted subjects with public identities. The checkpoint mechanism, which actually violates the core tenets of decentralization and permissionlessness.

In some cases, proof-of-stake systems have stronger security properties than proof-of-work systems. Because the value of the entire system is known (how many Tokens there are in the entire network, and the prices of Tokens are publicly known), if most of the "values" are agreed, then the number of participants is not so relevant. In this way, the network The problem of not knowing the number of all participants can be bypassed.

But to make sure they agree, everyone has to be online, and that's a strong security assumption, and Each block must be signed by thousands of nodes, which can have a large negative impact on performance. Therefore, although the number of participants is known, this does not compensate for the shortcomings of the proof-of-stake mechanism relative to the workload mechanism.

It can be seen that the security of proof of equity actually depends on many uncertain security assumptions, which make the whole system have a lot of Unstable Security Risk!

If you are interested in learning more about Mr. Ren’s thoughts, you are welcome to view the full speech video: https://www.youtube.com/watch?v=gxFm1QieUdE

Daniel: This article by Vitalik avoids the important and ignores the important

Daniel recently In an interview with the vernacular blockchain, he also expressed his views on Vatalik's new article:

The analysis and comparison of the advantages and disadvantages of PoW and PoS seem to be neglected in this article, and even distorted the facts.

The dispute between PoS and PoW, whether it is the blockchain technology community or the academic circle, They have never concluded that PoS is better than PoW, but they are divided into two camps. Everyone in their respective camps continues to dig deeper and find more suitable scenarios to give full play to their advantages and make use of their strengths and avoid their weaknesses. But when I read this article, I didn’t see any new research results or new evidence, but rather evasive assertions and imprecise conclusions.

1. PoS provides higher security at the same cost?

The calculation method of PoW attack cost in the article is almost completely unrealistic. The actual situation is that in the real world, only through The GPU computing power leased by the nicehash platform can launch attacks, but the GPU computing power has protected Ethereum well in the past five years and helped the total assets of the Ethereum platform to grow continuously. It has exceeded 100 billion U.S. dollars, isn’t it? contradiction?

Renting computing power through Nice Hash is a known way to launch a 51% attack on the network, and Nicehash can The total rented computing power is limited, and the number of PoW networks that can be attacked is also very limited, and it is meaningless to use professional mining machines to launch 51% attacks.

So it is necessary to distinguish whether it is safe to launch a 51% attack by renting GPU computing power and PoW, yes There are two different levels of problems. Being able to rent computing power through Nice Hash to launch attacks on small networks does not mean that the security of PoW is not enough. 51% attack is a deterministic problem that can be prevented and solved through mechanism design and plan, while the uncertainty contained in the PoS mechanism has more problems that cannot be solved.

2. Regarding the efficiency of recovery from attacks?

The article points out an attack against PoW, but does not talk about the premise. The premise of this kind of attack is that the network PoW algorithm is based on GPU, and the basic computing power is weak. You can rent enough GPU through nicehash to launch the attack. Only when all these conditions are met can there be an attack opportunity. This attack method is very harmful to ASIC The system is meaningless.

On the contrary, when PoS is attacked, as long as the attacker continues to attack, the performance is the same regardless of PoW or PoS, and even the PoS system may Worse, the slashing mechanism is of no help to solve the problem, and the attacker can completely profit from the outside, such as shorting the exchange, so what is the point of talking about PoS recovering more efficiently from the attack? Can the UASF mechanism of PoS, a mechanism to solve problems through human coordination, be used to prove that it is better than PoW?

3. Is PoS more decentralized than ASIC?

The understanding of decentralization, at least three levels, fairness, permissionless access mechanism, anti-censorship. At these three levels, PoS cannot produce any evidence to prove that it is even as good as PoW, so in the competition for the feature of decentralization, PoS does not compare with PoW, but chooses to compare with ASIC? However, ASIC did not make any damage on these three levels. On the other hand, PoS, the initial allocation problem, cannot achieve real permissionless access and anti-censorship, these are the problems of the PoS mechanism itself.

The closed nature of PoS causes far greater problems than imagined, and PoW borrows resources from the physical world to Maintain the consensus of the network, and the distribution of resources in the physical world is naturally decentralized. Physical resources are the most difficult to centralize, and the incentive mechanism encourages participants to continuously search for and mine lower cost and more efficient resources. This openness The traditional competition mechanism further leads to the dispersion of resources rather than concentration. In contrast to PoS, if a monopolist wants to maintain its monopoly advantage, then the cost he pays must be lower than that of competitors. The direct result of this mechanism is system involution.

About the problem of weak subjectivity, this problem does not exist in the PoW system, only in PoS, when judging the main branch, there is no Avoid reliance on external information. This problem is unsolvable in the PoS system, and it is also a problem that can lead to far greater problems than imagined. In the early PoS research of Ethereum, the concept of "weak subjectivity" was created by itself, but this concept has not really been recognized by the academic community until today.

Cipher: PoS is suitable as a transaction layer rather than a value storage layer

Nervos researcher Cipher also published an article last year: Mining to the left, Staking to the right——Two major evolution directions and value analysis of the public chain

https://talk.nervos.org/t/mining-staking/3028

Cipher believes that PoS has three core problems, weak subjectivity, security and business coupling, and censorship, which is very suitable for use as a transaction layer rather than a value storage layer :

The public chain can be divided into two categories. The public chain of hosting services , and the second type is the public chain whose main performance will be positioned in the decentralized transaction of assets. And he believes that the final choice of the first type of public chain is PoW, while the second type of public chain will choose PoS in order to emphasize performance and efficiency.

The strong decentralization of the first type of public chain and the compatibility of the second type of public chain with real world assets are both society needs. However, there are obvious differences in the positioning of the two types of public chains, targeting different asset types and business types. A public chain must choose a path with a clear-cut banner and go down firmly. If it swings left and right, it will lose users on both sides. The two can be connected in a cross-chain manner, and users can transfer assets on the two types of public chains according to their own needs. This is also an inevitable trend in the development of public chains in the future.

This is also the focus that Nervos has emphasized for a long time. We don’t think there is any problem with PoS, but the use of PoS The occasion is not the underlying value storage or settlement guarantee. At the same time, regarding Vitalik's article, Cipher also believes that PoS has three problems:

1. Weak subjectivity, judging the main branch needs to rely on external information;

2. Business and security coupling, for example, when the DeFi interest rate is higher than the staking interest rate, the system security is reduced;

3.Non -permissionless mining, anyone who participates in mining must buy stake, but stake holders can not sell, causing the system to roll.

How do Bitcoin and Ethereum core communities view PoW and PoS?

Of course, although most of the current public chain development They all chose PoS, which is cold-started and easier to operate. However, many bigwigs in the blockchain circle have expressed concerns about PoS. On the day when Vitalik’s article was published, Ryan Gentry of the Bitcoin Lightning Network Lab also had some wonderful conversations and exchange of opinions with Vitalik; Mr. A Jian of Ethfans, a Chinese Ethereum and blockchain technology community, also wrote an article expressing his opinion on “POS + sharding" path.

A Jian believes that there are too many assumptions about performance and security in the road of "PoS + sharding". A summary of the route changes of Ethereum in recent years: "Today, there is no evidence that Ethereum should embrace PoS for scalability", "There is no roadmap related to ETH 2.0 worth Ethereum Square abandons the current mode of operation and turns to a PoS-based system.”

For more information, you can refer to what Mr. A Jian posted yesterday: view| Fang's route Should the graph change?

Ryan Gentry is a core member of Lightning Labs who specializes in Lightning Network. He commented on Twitter after seeing Vitalik's article:

(Vitalik 's article) lacks analysis of key PoS errors. The PoS consensus has basically been captured by exchanges and custodians for free (note: it means that they can obtain a lot of ETH for Staking at zero cost). The only consensus cryptocurrency that can be captured is fiat currency. However, the equation of PoS = legal currency is equivalent to the "long-term" argument that Keynes said (Keynes said: I only consider short-term issues. In the long-term, we are all dead)

Vitalik also responded:

This is in Exchanges "would" stake, but even if they did, it certainly wouldn't involve all of their customer balances. However PoW may be acquired by large mining companies.

Ryan continues to respond to Vitalik:

Why don't the exchanges use customers' funds as bets? Are exchanges notorious for leaving money on the table? We have empirical evidence to the contrary. And PoW is decentralized because even cheap energy is decentralized. This successful example happened to Bitcoin today.

Vitalik then responded:

Hardware cost is the main cost of PoW; therefore, if the sector is dominated by large miner manufacturers, it does not matter what the energy situation is.

In addition, Vitalik then posted another tweet in response:

Regarding the question "why don't they bet customers' funds?", I think it is illegal to use users' funds without their explicit consent. Withdrawal delays, risk reduction, leaks and penalties, etc. (Note: These are all things that need to be considered by centralized exchanges). Even if some colluding organizations or small groups really obtain 51% of the Tokens in the entire network, if they do any evil in the community, they still have to run the risk that the community may carry out soft forks through "inactivity leak" . They also have nothing bad to do.

Ryan thinks this argument is untenable:

That's ridiculous. So, the security of ETH2 relies on a large exchange adhering to laws that do not currently exist, and that will exist after consensus is reached? This is much weaker than globally distributed energy and out-of-protocol ASICs.

Based on the above arguments, we can know that so far, at the bottom In the blockchain, the PoS consensus has not yet had a relatively successful case. One reason is because of the development time, but the other reason is also because of the consensus of the reward and punishment system, the difficulty in mechanism design must be greater. At the same time, in the case of different acquisition costs of mortgage tokens for each mortgager, even centralized exchanges and custodians are considered to be able to embezzle other people's funds for mortgage without compensation. This point of contention also seems to be difficult to have convincing arguments. At the same time, whether the punishment mechanism may also affect the innocent has not yet had a good answer.

Finally, with the growth of the Ethereum ecosystem. The state on the ETH chain is also increasing day by day, making it increasingly difficult for many developers to maintain their own nodes. Therefore, the dependence on Infura and so on is becoming stronger day by day. On the eve of ETH 2.0, the state on the ETH blockchain will only continue to grow. How many people and how many teams can synchronize nodes online at all times in the ideal of ETH 2.0 is also a very big problem. The big unknown.

The above PoS questions were not raised on the first day, and there has never been a perfect answer. We may often hear a lot of arguments that "PoW is too wasteful of electricity", but reaching a consensus will inevitably require a cost, not to mention the consensus of the entire network. PoW is the most direct way to reflect the cost on electricity. PoS is not without cost, but shifts the cost to "security risks", "mobility of funds" and "currently endless coordination and mechanism design". This is The PoS consensus still needs to be paid as a Layer 1 cost.

However, as Cipher said, for Nervos, PoS is not bad, but should be used as an asset storage When developing the underlying blockchain, it is bound to face more stringent security inspections and challenges. Of course, PoS is still very suitable as the transaction layer consensus of Layer 2, providing performance requirements that Layer 1 cannot meet. The reward and punishment mechanism for pledged funds has been discussed in today's side chains and Rollups.

Of course, we still expect PoS to overcome many obstacles in the future and find a blockchain track suitable for PoS development, so that the entire blockchain ecology can develop more vigorously.