Author’s Note: For readers who are already quite familiar with EIP-1559, you can go straight to the fourth subsection of the third section below. In this section, I provide a fairly complete analysis. You can even know from this, which arguments put forward by the supporters and under which circumstances are actually correct. However, none of them could describe the full impact.
Readers who are not familiar with this subject can start from the beginning. If you need an introduction to the content of EIP-1559, please see.
I. Introduction
The name “EIP-1559” must be familiar to everyone. This proposal, which was formed in March 2019, introduced a mechanism to destroy handling fees (thus reducing the supply), and was raised by David Hoffman as the “last piece of the puzzle in Ethereum’s monetary policy”. It was that ETH “obtained a currency premium.” “The key part of the world” has received more and more attention.
A year and a half ago, after reading the introductory article written by Eric Conner, the second author of EIP-1559 proposal, I wrote a rigorously worded article, thinking that this proposal does not solve the problem that the proposal originally intended to solve. Introduce more transaction friction. My friend Elisa also helped me to translate the article into English, but unfortunately there were few people who responded.
Time has changed, and voices supporting EIP-1559 have repeatedly appeared recently. Whether it is the expectation of “deflationary Ethereum” or the anxiety caused by the rise of Gas Price, it seems that it has increased the popularity of support arguments.
But I never saw a reason that could convince me. These supporting opinions have either misunderstood the operating principle of the fee market (it is actually and can only be a so-called “one-price auction”, where the owner can set a reserve price and the higher price will get the auction); or only see the EIP -1559 will cause part of the impact, while turning a blind eye to its other impacts; in short, they can’t provide a complete analysis.
I have collected several recent articles that support EIP-1559:
- Analysis of EIP-1559 (Chinese version)
- What if ETH had a fee burn 5 years ago
- EIP-1559 51% Attacks: Should you live in fear()
- Ethereum fee market reform: EIP-1559 as a question of fairness()
I very much respect the intellectual resources invested by the authors of these articles. In any case, without these analyses, the whole picture would not become clearer. But in my opinion, they were a bit sloppy when drawing conclusions, lacking arguments for some key claims, or proving that the new mechanism cannot be manipulated, but they did not prove that the new mechanism would be better than the current mechanism (below We will use “#” and numbers to refer to these articles). I start with some simple economic reasoning, and then discuss in detail the problems in these articles.
2. Simple economic reasoning
Assuming that the market price of an item is very high, do you think that depriving the producer (seller) of the economic income from this item can make the cost of obtaining this item lower? Obviously not, because the price is determined by supply and demand, and cutting the income of sellers will dampen their enthusiasm for production, and it will also block the signal effect of price (margin), allowing potential producers to enter this production field. The decline in enthusiasm affects the extent of the future increase in supply. The supply can’t be increased, and it’s useless to say more. But EIP-1559 just wants you to believe that cutting the income miners get from gas fees can bring down the gas price. how can that be possible? Suppose you go to the hospital and find that the doctor’s outpatient fee is very high. You think it is expensive. Which of the following methods do you think will enable you to get a more cost-effective service (the same price can get better service, or the same service only Need to pay less)? A. Lobbying the government to introduce control measures for outpatient fees, classifying doctors, and delimiting the upper limit of outpatient fees for each level of doctors; B. Complaining doctors on social media and scolding doctors for profit, regardless of whether the patient lives or lives. , It’s not ethical at all; C. The government starts to collect taxes on outpatient fees, or increase the tax rate of outpatient fees; D. While the government collects taxes on outpatient fees, it comforts everyone that it’s okay. I put this money in an open account and will never use it to help you deflate; E. The government regulates the price of doctors’ consultation fees, and at the same time provides doctors with subsidies, and finally taxes on consultation fees; F. The outpatient payment is taxed. On the one hand, these taxes will become a pool of funds, which will enjoy the interest of bank savings. After a period of time, the funds will be returned to the doctor in proportion to the cost of the consultation. Which method do you think is useful? None of them are useful. #A, controlling the price of consultation fees (obviously compulsory reduction of consultation fees) will lead to a large increase in demand and insufficient supply. This is reflected in the upper limit of the number of registered numbers, the increase in queuing, and the scalpers who help register . In addition, the average consultation time for patients will also decrease. Do you think you really got the benefits? No, although your monetary cost has dropped, the cost of time (queuing) has risen. Furthermore, it causes patients to be unable to use bids to express their urgency for medical resources, and the actual use of medical resources becomes more time-valued. Low group excursion. According to the rhetoric in the industry, high-quality medical resources have been attacked by DoS. #B, this need not be explained, right? It is equivalent to imposing a negative price on the compensation of doctors. #C, for the same consultation, the income that can be obtained is lower, and the doctors are willing to spend less. Moreover, the wool comes from the sheep, and all taxes are squeezed from the buyer and the seller at the same time, so that when both parties are insufficient, the transaction will not happen at all. #D, the discerning person should have noticed, in fact, this is EIP-1559. It is actually #C (the effect of tax) plus the effect of currency deflation. Currency deflation certainly has the effect of causing currency value appreciation, but don’t be fooled by the appearance: If destroying part of your own property can make your situation better, why don’t we see people often burn their own money? (Assuming that the amount of deflation will increase the value of money proportionally (this assumption is of course unreasonable), this is actually a simple mathematical problem) In the combination of “tax + deflation”, the price of doctor’s consultation fee (in terms of monetary quantity) Said) become lower, but the value of the currency rises. Even if no detailed analysis is required, it should be said that “one high and one low is hard to say”, but if this can increase their income, why have we not seen the consumption tax? Industry groups lobbied the government to destroy the tax? #E, some people argue that #E is the actual situation of EIP-1559, because miners not only have gas fee income, but also block rewards. But in fact, there is no difference, because block rewards can only motivate them to provide proof of work, not gas. It is like changing the income structure of doctors and replacing medical fees with government subsidies. The result is to encourage doctors to sit on duty, not to encourage doctors to visit. I don’t know if you have ever heard of doctors in some primary hospitals who are doing referrals and never actually see them? There is only one logical key to link EIP-1559 with the above example, that is, EIP-1559 is first a tax. Why do you say that? The key measure of EIP-1559 is to make the amount paid by the user (buyer) not equal to the amount received by the miner (seller). The total amount paid is divided into base fee and tip (the amount actually received by the miner). The user spent 10 yuan, but the miner only got 5 yuan. Where did the middle 5 yuan go? Don’t care where it goes, isn’t it just a consumption tax!
– Source:-As long as you know that it is a consumption tax with a specific amount, you can think of the two most basic principles in tax economics: (1) A consumption tax is never a benefit only collected from producers, but also Benefits are collected from consumers. Looking at the green part of the picture above, it’s clear at a glance. Originally, the parts above P1 and below the blue demand curve are the benefits that consumers get, that is, they are willing to pay higher prices (“Liquidity mining daily chemical yield 100% , Operating a wave”), and only paid a lower price (“Gas fee is only 1%, which is okay”), and thus obtained a part of the benefit called “consumer surplus”; similarly, there are “producers Surplus”; but taxation does not only encroach on producer surplus or consumer surplus, but both; (2) Where there is a consumption tax, there must be inefficient losses, that is, the surplus of the buyer and seller is not enough to pay the tax, so the transaction It doesn’t happen at all. It is the triangle formed by the angle between the supply curve and the demand curve on the right side of the green square in the figure above. That is, EIP-1559 will cause both the supply and consumption of Gas to decline. Many people claim that EIP-1559 can improve the user experience, but this is obviously not analyzed. After analyzing it, I don’t understand how to say that the user experience will be better. At least, the amount users have to pay has not decreased. As for the deflation caused by it, I have already mentioned it. Deflation, of course, causes an increase in the value of money, but since it has no reason to prove that the income of miners from Gas will become higher, there is no way to prove that miners are more willing to optimize the network and nodes and provide more Gas (that is, the gas supply). Will increase), naturally there is no way to prove that it can reduce the gas price. What’s more, it gained this deflation at the expense of ETH’s property attributes. Next, I will discuss the fallacies of the above articles.
Three. The Wise Men
In this part, I will respond to several basic concepts and theories in the argumentation of the above four articles. These concepts mainly come from #1, #3 and #4. This is of course because #2 does not actually provide any argument, only extracts the historical Gas consumption and Gas Price data, crudely assumes a destruction ratio and calculates the destruction amount. However, a blockchain that has implemented EIP-1559 cannot have the same smoothness and transaction volume as a blockchain that has not been implemented.
(1) #1 and Slack mechanism
#1 It is argued in the article that EIP-1559 makes the actual size of the block more flexible and able to cope with sudden surges in demand. In this way, some blocks can be larger without becoming too large as soon as they are expanded. Long-term burden. This proposition is reasonable. EIP-1559 defines two concepts related to gas capacity: target gas capacity and maximum gas capacity; the former and actual gas usage are used to determine whether the base fee rate should be increased or decreased, and the latter is the maximum capacity of a block. The amount of gas used. Therefore, when the demand suddenly increases, miners can pack larger blocks in a short period of time without going through a lengthy consensus formation process. I did not mention this in the last article, it was my negligence. But this is not without cost. Assuming a sudden drop in demand and the rate that users are willing to pay directly falls below the base fee rate, the Ethereum network has to empty blocks and wait for the base fee to adjust downward. Up.
(2) #3 and the security of EIP-1559
#3 Dedicated to prove that EIP-1559 is not controllable, even if it is a 51% attack on proof of work, EIP-1559 cannot be manipulated. In my opinion, the argument that EIP-1559 is not controllable is completely wrong, or it is claiming a benefit that is not exclusive to itself. Because miners don’t need to manipulate it at all. It turns out that how miners make money, after the implementation of EIP-1559, they still make money, but the name is changed, the original earning is called Gas Fee, and now the earning is called Tip. Although the names are different, they are actually the same. As long as miners still have the power to pack blocks, you cannot force them to pack transactions that will make them lose money (that is, transactions whose tip is not enough to cover the opportunity cost they paid for packaging this transaction), they can do it themselves Determine the amount of gas supplied. Don’t get me wrong, I mean, I also think EIP-1559 is not controllable, and the cost of manipulating it is greater than the gain. But because miners don’t need to manipulate it at all, this benefit doesn’t exist.
(3) #4 and “Commons”
I must admit that the most shocking thing to me is the “commons” doctrine proposed by #4. In the section “Who owns Ethereum’s blockspace”, the author advocates that the work of miners to protect the network has been bought out by block rewards, plus processing The cost of the transaction is not only borne by the miners, so miners cannot claim that they “own” Ethereum blocks, and naturally they cannot claim that they should receive transaction fees. The author’s conclusion is: This means that the block space of Ethereum is the “commons” and the miners are “rent seekers on the commons”. It really surprised me. What surprises me is not only that the author seems to have completely misunderstood the role of Proof of Work (PoW) and overlooked the core function of PoW in ordering transactions in a distributed system. It is precisely because PoW allows us to sequence transactions in a distributed system that we need it. This sorting function is essentially the same as what the author calls “securing the network”. From an individual perspective, it is true that miners provide proof of work to obtain block rewards; but from a network perspective, packaged transactions and transaction sequencing are the core functions of PoW. If this function is missing (whether it is banned from the agreement, or the incentive to package transactions is eliminated), the harm is the agreement itself, or it will become a centralized system, or it will harm the property of the agreement currency Attributes (property that cannot be transferred is worthless), and another possibility is that users must pay for this price in other ways than currencies (very poor efficiency) (Do you want to guess which item?). On the other hand, what surprises me is that the author is so close to the truth. Yes, if we agree with the author’s reasoning that the block space of Ethereum is a common land, then the author should also think that the first principle of common land governance is privatization, whether directly or indirectly. In our case, we use competitive advantage to define property rights, that is, according to the ability of collection and package transactions to determine which mining pool can get how much share and how much revenue. The author said that miners are rent-seekers. I said, of course, they are rent-seekers. Can they be anything else? Everyone, as long as they pursue “producer surplus”, is by definition rent seekers. The important thing is that the world is booming, all for profit, and the world is bustling, all for profit. If you don’t let them enjoy the rent, the land will be deserted. End users will get even less.
(4) #1, #4 and “Improve Security”
In #1 and #4, the author mentioned an argument: when transaction fees become the main part of miners’ income, because transaction fees will fluctuate greatly, the computing power input of profit-seeking miners will also fluctuate. Security will also fluctuate, so the Ethereum network is in an unfavorable situation; EIP-1559 can eliminate this risk once and for all, by reducing miners’ fee income while maintaining the value of block rewards to maintain the stability of miners’ computing power input Sex. Whether the supporters of EIP-1559 realize it or not, this is actually the most complicated and ultimate argument for one party. It combines the Ethereum community’s understanding of network security policy (that is, monetary policy) and finds an ultimate benefit. However, as long as we use the tools we introduced above (supply curve, demand curve, market equilibrium, producer surplus, consumer surplus), we can also analyze whether this argument is correct. Because in the final analysis, what we have to do is to judge the changes in miners’ income. If the revenue increases, then the computing power to maintain the network will increase. The concept of “residual” represents the benefits of transaction participants. Under the current fee mechanism, only two pictures can cover all market conditions, namely:
The orange supply curve represents the miner’s supply curve. This curve means that for every additional unit of gas miners provide, the required marginal price rises, because providing gas means paying for the amount of calculation and time, and the assembly area The more time a block takes, the greater the risk of its block becoming an uncle. The blue demand curve means that the user’s bid for each additional unit of Gas is decreasing, because users will put their urgent needs ahead. The intersection of the two is called the “equilibrium point”, which means that the user’s willingness to bid is consistent with the miner’s willingness to ask. The price at this point is the Gas Price in the market, and the amount of use that this point means is the actual amount of use. In the two figures, the supply curve has a vertical upward part. This is because the gas supply of the miners is restricted by the agreement. When the supply reaches the upper limit of the block gas specified in the agreement, no matter how the bid is made, the miner cannot provide more change. too much. The picture on the left describes the general situation, or the situation where the market demand is low. At this time, the gas usage of the market equilibrium does not exceed the block Gas upper limit; while the picture on the right describes the situation where market demand is relatively strong, and users want There are a lot of gas, but miners can only provide that much because of the agreement, so the gas usage is equal to the block gas upper limit. This shows that when we assume that the supply curve does not change, the “equilibrium point” can be a useful tool for describing market demand. After joining EIP-1559, the operation of the gas market can be divided into 5 types: A. The natural market equilibrium point is greater than target gas usage and less than max gas usage (ie 2 times the target usage); base fee is zero; B. Natural market The equilibrium point is greater than the target gas usage and less than the maximum gas usage; the beae fee is not zero, but not so large that the gas usage equals the target usage; C. The natural market equilibrium point is greater than the target gas usage and less than the maximum gas usage; the base fee is not Zero, and large enough to make the actual gas usage of the network equal to the target usage; D. The natural market equilibrium point is less than the target Gas usage, and the base fee is not zero; E. The natural market equilibrium point is less than the target Gas usage, and the base fee is zero ; Readers should be aware that: (I) corresponds to a sudden surge in demand; (II) corresponds to a situation in which the base fee is being adjusted (increased) after the increase in demand; (III) corresponds to the completion of the adjustment of the base fee to increase network usage The situation is approaching the target usage; (IV) corresponds to the situation where the base fee is being adjusted (decreased) after the demand is reduced; (V) corresponds to the situation where the base fee adjustment is completed and becomes 0 after the demand is reduced. As long as two assumptions are added, we can analyze whether EIP-1559 (compared with the non-implemented situation) has created more income for miners in the above five scenarios: (1) Hypothesis 1: EIP-1559 The target gas consumption of the project is equal to the upper gas limit of the block when not implemented; (2) Hypothesis 2: The surplus drawn from miners and users by EIP-1559 will be losslessly converted into the value increase of the block reward (this is an obvious Unrealistic hypothesis, but it is convenient for reasoning; and after the reasoning is completed, if you give up this hypothesis, the conclusion will be obvious) Take the situation of #A as an example, the situation at that time can be described by the following diagram:
If EIP-1559 is not implemented, the gas supply will be restricted by the agreement, so the gas usage is GL (block gas upper limit), and the gas price paid by users is P1 (note: the price is not the miner’s supply curve) The level of buckling that began to appear is because of competition among Gas consumers). The income of miners is area B + area C + area D + block rewards. Consumer surplus is area A. If we implement 1559, because the mechanism allows the block to be temporarily expanded (Hasu’s so-called “Slack Mechanism”), and the base fee is zero at this time, the gas usage is U and the gas price is P2. At this time, the consumer surplus is area A + area B + area E, which is significantly larger than area A. This is the benefit Slack Mechansim brings to consumers. But at this time, the income of miners is area C + area D + area F + block reward. Compared with area F, which is bigger and smaller? Under the current mechanism, miners can collectively increase the block Gas upper limit, so if area F is greater than area B, and the coordination cost is low enough, miners will choose to increase the block Gas upper limit. Therefore, we can basically conclude that under such market conditions, the implementation of EIP-1559 will reduce the income of miners. Using the same method, we can also analyze #B and #C:
When the base fee starts to rise, but it is not high enough to make the gas usage of the network reach the target usage (ie #B), take the TIP level and TIP+BF level indicated by the above icon as an example. At this time, the supply curve (including the dotted line) gives the amount of tip that must be paid for the miner to pack the transaction, but the base fee is not zero. Under the interaction of the two prices, although the user pays tip+base fee, The miner only got the tip, and the difference in between was the tax that was withdrawn. Similarly, if EIP-1559 is not implemented, the miner’s income is area B + area C + area D + area E + block reward, and if it is implemented, the miner will get area D + area E + area H + block reward*. The reason why it is a block reward* rather than a block reward is because the tax evacuated by the base fee mechanism will also increase the value of the block reward through burning. According to our hypothesis 2, this means that the miner will actually get Area C + Area D + Area E + Area G + Area H + Block rewards. Consumers will get area A + area B + area F, mainly because they benefit from a larger block space (assuming that the block space is expanded to U, even if EIP-1559 is not implemented, they can get the same Many benefits). As for whether the income of miners has increased or decreased, it depends on which area B and (area G + area H) is larger. (According to my personal opinion, I think B is larger, because under the current mechanism, if the coordination cost is low, GL is the block size that can maximize the income of miners.) Similarly, if it is #C, it means base fee If the gas usage is so high that the gas usage is equal to the target gas usage (that is, the upper limit of gas when not implemented), the miner’s income will become area E + block reward*. As for area B + area C + area D, both are The base fee (tax system) takes away the surplus; according to Hypothesis 2, the actual benefits of miners will become area B + area C + area D + area E + block rewards, yes, it happens to be the same as when EIP-1559 is not implemented . The benefits of consumers will also change back to area A, which is the same as the benefits under the current fee mechanism. As for #D and #E, as shown below:
Assuming that U is less than the block Gas upper limit, under the current fee mechanism, the miner’s benefit is area C + area D + area E + block reward, and the user’s benefit is area A + area B + area F. After implementing EIP-1559, assuming that the base fee has not yet dropped to 0, the area B + area C is the amount burned because of the base fee. The income of the miner will become area D + block reward*, that is, area B + area C + area D + block reward. Whether it is increased or decreased depends on the relative size of area B and area E. However, the user’s benefit becomes area A, which is the other side of the Slack Mechanism coin. It is easy to see that area F + area E can be realized under the current mechanism, and after the implementation of EIP-1559, the benefits that cannot be realized because of the base fee, that is, as mentioned above, “net tax loss “. If the base fee drops to zero, the benefits of miners and users are the same as under the current fee mechanism. In summary, we can draw several conclusions:
- When demand soars, because EIP-1559 allows miners to produce blocks that are twice as large, consumers can get more benefits. But when the market demand drops, but the base fee has not yet dropped to zero, consumers must pay more expensive fees than the current fee mechanism.
- EIP-1559 does not necessarily bring higher security. At least, in #C and #E, the security will be equivalent to the current mechanism.
- However, #C and #E are two static equilibrium points, that is, in reality, we either can’t just reach it, or we can’t maintain it for a long time. So we should pay more attention to several other states.
- On the surface, in #B, there is no way to assert that EIP-1559 will reduce the income of miners. But in fact, under the current mechanism, if miners find that the existing block Gas upper limit cannot help them maximize transaction fee income, they can increase the block Gas upper limit. This also means that a larger block (EIP-1559) actually cannot help them increase their income. The only situation that we cannot infer further is #D. At this time, we really don’t know which area B and area E is larger. But at least, the user’s revenue is clearly reduced.
- All the above conclusions rely on Hypothesis 2. If we relax Hypothesis 2 (in fact, we should also relax it, because even according to the rough quantity theory of money, deflation will increase the value of each unit of currency, and it is impossible to increase the value of the newly issued currency), we can There is great certainty: EIP-1559 will reduce the security of Ethereum.
If we really want to implement the “minimum issuance rate” policy, we should not implement EIP-1559.
Four. Conclusion
In summary, I have proved that the current support for EIP-1559 does not have a solid scientific basis. EIP-1559 can neither make transaction fees more predictable (because demand is unpredictable), nor can it reduce Gas prices, so it cannot be said to create a better user experience. At the same time, the newly emerging supportive opinions either misunderstood the operating principles of the entire ecosystem or failed to provide a complete analysis to prove their opinions. So far, no core developer has expressed a plan to implement EIP-1559 on the PoW Ethereum, but the 2.0 researcher said that a similar mechanism may be implemented in Phase 1. What I am more concerned about is whether PoW Ethereum will decide to implement this mechanism because of the current general support. If these opinions are only because the authors have not clarified the cause and effect, it is still easy to say. What I worry more about is that someone who knows its impact and chooses to support such an unfair proposal because he does not belong to a group that will be harmed. Again, I have great respect for all those who invest intellectual resources on this topic. Without their voices, it would be impossible for me to advance my own analysis and write this article. I have also repeatedly emphasized that these words should have been said by some people, and they do not need any talents. However, no one said, so I had to say it myself. Finally, a private point of view is entrained. Since the history of mankind, there has not been a system that can stimulate people to engage in production more than a market economy, and no system can reduce costs more than a market economy, because other systems induce people to profit from non-productive activities. The blockchain world is no exception. If you look at the block explorer, half a year ago, the block utilization rate of the Ethereum network was still around 80%, and in the past month, it has been above 95%. why? When the transaction fee is too low, many mining pools would rather pack empty blocks. But when the Gas Price soars, 0.0x or even 0.x less ETH will be lost without a single transaction. Those mining pools that have not optimized nodes and networks will face the tragic end of watching others eat meat and being abandoned by miners.
This fact points us to the right path.
