Wu's in-depth talk: 7 major changes after Ethereum 1559 upgrade: Who can achieve it and who will fail

Wu Shuo Author | Wu Zhuocheng

Editor of this issue | Colin Wu

1 Design Principle

The essence of blockchain is a series of ledgers. In a decentralized world, no one is obliged to keep accounts for users. The work of keeping accounts and ledgers needs to be completed by miners who participate in mining. Mining is not only about keeping accounts, but also about ensuring the security of the ecological chain. A network with insufficient computing power is extremely vulnerable to attacks by 51% computing power, and security issues will be threatened. Therefore, the user's payment mechanism and the miner's income mechanism determine the efficiency and security of the entire block network. When users pay too much, the number of users participating in on-chain transactions will decrease, resulting in a waste of block space and reducing the value of the entire network; when miners' income is too low, the computing power invested in mining will decrease, and the security of the network will be difficult to guarantee.

Therefore, the setting of transaction fees is a process of balancing supply and demand. Miners are the suppliers, users are the demanders, and commodities are objects called Gas in the Ethereum network, which can be understood as the amount of work required for each transaction. For example, the most basic work, a simple transfer, costs 21,000 units of Gas, and more complex work, such as the execution of smart contracts, requires more Gas. Of course, the size of Gas that a block can accommodate is not unlimited. This limit is called the Gas limit and is voted on by miners. The most recent adjustment was proposed by Vitalik on April 23. Although it was opposed by some miners, it was eventually increased from 12.5 million to 15 million, thereby alleviating the network congestion caused by the increase in on-chain activities.

Here we need to explain why miners object, which helps to understand the role of the 1559 protocol. Before each accounting, the blockchain network will publish an encrypted block. After receiving the message of the block release, the miners will perform algorithm decryption. The miner who first solves the puzzle will obtain the right to record the block, and the Ethereum algorithm will give a fixed block reward for this. So how do miners record accounts? In abstract terms, the miners who have mined this block "pack" some information waiting for confirmation in the memory pool and fill it into the block, and then put it into the blockchain network for transmission. Therefore, the smaller the block capacity, the less information is packed each time. Scarcity makes things valuable. In order to be packed as soon as possible, users must naturally pay higher rewards, and miners can profit from it. This is why miners are reluctant to increase the Gas limit.

Here we introduce a third concept, namely Gas price, which represents the fee that users are willing to pay for each unit of Gas. The first-price auction principle is adopted, that is, the highest bidder wins, from which the total fee can be calculated to be equal to Gas×Gas price. Throughout the process, miners can obtain block rewards provided by the algorithm and transaction fees given by users. Among them, block rewards are fixed (the protocol can be changed by voting), and transaction fees have a large floating space. With the increasing number of users on the Ethereum network, Gas fees continue to rise. In September last year, Gas price once rose to 478Gwei, which means that when the block utilization rate is 100%, the transaction fee is nearly 6 ETH (12.5 million×478Gwei÷1 billion). The block reward is only 2 ETH, so transaction fees have become the main income of miners.

Since miners tend to create scarcity, why did the proposal pass in the end? The reason is that with the surge in Ethereum users, block space is in serious short supply, so that even if the Gas limit is increased, users still need to bid high prices to compete for limited space. The supply and demand curve in the figure below can well explain this phenomenon:

Among them, D1 is the Gas demand curve before expansion, and D2 is the demand curve after expansion; the supply line is the same. It is not difficult to see that with the increase of the Gas limit, the demand curve will shift to the left under the premise that the number of users has not increased, and the Gas fee will decrease accordingly, but due to the increase in the amount of Gas, the transaction fee received by miners will not decrease. Not only that, in fact, due to the continuous increase in the number of users, the user demand for Gas has not decreased due to the increase in the Gas limit, so the Gas price has increased instead of decreased.

It should be noted that the supply involved here (and in the following text) refers to the amount of work that miners are willing to do to package information, not the size of the block. The block size is fixed to the Gas limit (actually slightly floating), and is not affected by the miner's computing power; similarly, the block time is only affected by the decryption difficulty. The Ethereum algorithm will automatically adjust the difficulty according to the computing power invested by the miners to achieve an average speed of one block in 13 seconds. Therefore, what is really controlled by the miners is the accounting power, that is, the miners can choose not to package any content and only pass an empty block out. This is also the biggest gaming advantage of miners behind the 1559 protocol.

2 What is the EIP-1559 protocol

The biggest change of EIP-1559 is that it splits the original transaction fee into base fee and tip. Among them, the base fee is the fee that users must pay, and the amount is determined by the utilization rate of the previous block, rather than by the user auction. To this end, the 1559 protocol also deliberately changed the upper limit of the block space, from the previous 15 million to 30 million, but still with 15 million as the target size. The intention of this design is to maintain the block utilization rate at 50% to facilitate the calculation of the base fee. When the block utilization rate exceeds 50%, the system will automatically increase the base fee; vice versa. The specific calculation is as follows:

Among them, b0 is the base fee of the previous block, and s0 is the utilization rate of the previous block. Through calculation, we can find that when there are 39 consecutive blocks with utilization rates close to 100%, the base fee will increase to 99 times the original (1.125^39). In other words, if the initial base fee is 10Gwei, then in less than nine minutes, this fee will exceed 1000Gwei. Obviously, users cannot afford such high transaction fees, so as the base fee increases, the amount of information on the chain will decrease. After several blocks with utilization rates of less than 50% appear, the base fee will also fall. In this way, the protocol can maintain the block in a "half-full" state.

During this process, the wallet will automatically calculate the current base fee based on the size of the previous block, and the user can choose to accept or not. If not accepted, the user can only wait for the base fee to drop before being packaged. This is not a bad idea if the transaction is not very urgent. If accepted, the information may be successfully packaged in the current block. Note that this is only a possibility, because the system calculates the base fee based on the utilization rate of the previous block, but it cannot predict the utilization rate of the current block. Assume a situation: Since the utilization rate of the previous blocks is less than 50%, the base fee has dropped significantly. At this time, users think that the fee is low, so they pay one after another, and even the amount of gas purchased exceeds the maximum capacity of the current block. At this time, some information cannot be packaged in the current block and can only wait for the next block. However, since the current block utilization rate reaches 100%, the base fee of the next block will increase. If no additional transaction fee is added, this part of the information can only continue to wait until the base fee drops again.

Therefore, if you want to ensure that the information is packaged immediately, the user must pay an additional fee, namely the tip, on top of the base fee. Unlike the base fee, the tip size is determined by the user himself, and the decision method is still the first-price auction, that is, the higher the tip, the more likely it is to be packaged immediately. In addition, the tip will be paid directly to the miner, while the base fee will be destroyed. For example, assuming that the base fee of the current block is 100Gwei, and the user sets the fee cap to 200Gwei+5Gwei, this means that the user is willing to pay a maximum of 200Gwei base fee (in case the current block cannot be packaged and the base fee of the next block increases) and is also willing to pay the miner 5Gwei tip, and the final fee paid is 105Gwei. Here we need to explain why the base fee must be destroyed. Otherwise, miners will conduct frequent small transactions on the network, resulting in high block utilization and exponential explosion of the base fee.

3 Impact on the Ethereum Network

The EIP-1559 protocol was proposed and designed by Vitalik Buterin and Eric Conner. Eric Conner elaborated on the motivation and advantages of the protocol on Medium, and believed that the transaction fee mechanism can bring the following benefits to the Ethereum network and its users:

a) Save up to 90% of transaction costs

b) Improve user experience by automating the transaction fee bidding system

c) Provide users with a predictable transaction fee system

d) Reduce the expected waiting time for transaction confirmation

e) Even if transaction fees are the main source of rewards, this approach can also discourage selfish mining

f) Allow users to “jump the queue” even when the network is congested

g) Consolidate the economic value of ETH at the protocol level

This article summarizes and summarizes what Eric Conner said. The first five points can be roughly summarized into the following two points:

a) Significantly reduce transaction fees

b) Significantly reduce the volatility of transaction fees and make transaction fees predictable

As for the last two points: allowing “queue jumping” has existed since ancient times, which is nothing more than continuously increasing the tip, which is no different from the current first-price auction; and increasing the value of ETH is a side benefit. If the first five points can be achieved, the value of the entire ecosystem will naturally increase. This point can be understood as the final manifestation of all the above advantages. The following article analyzes the two summarized visions to explore their feasibility.

3.1 Whether to significantly reduce transaction fees

Through the interpretation of the transaction fee mechanism, it is not difficult to find that when transactions are congested, the tip under the 1559 protocol is actually no different from the existing transaction fee. Users with urgent transaction needs must pay tips to compete for limited space. At this time, the protocol is equivalent to returning to the existing first-price auction model. It’s just that a part of the entry fee, the base fee, must be paid before the auction. Therefore, the design of this fee is equivalent to adding a tax on the basis of the existing fee. In the centralized world, taxes belong to the state, but in the blockchain world, since there is no central agency, this part of the fee can only be destroyed.

In the first section of the article, when explaining the design principle of the transaction fee mechanism, it was mentioned that the trade-off between miners and users is actually a game between producers and consumers, and the commodity being bought and sold is the workload spent on bookkeeping, which we call Gas. At this point, the discussion of the impact of base fee on Gas price can be transformed into an analysis of the impact of taxation on price, and the research method of economics is applied here.

As shown in the figure above, we will not consider the shift of the demand curve caused by the change in the number of users. Assuming that the number of users remains unchanged, the demand curve D will remain unchanged. Here we need to explain that under the 1559 protocol, the supply and demand curves cannot intersect the vertical part of the supply line for a long time, because that means that the block utilization rate is always at 100%, which will cause the base fee to rise exponentially. Therefore, in the long run, the equilibrium point of the supply and demand curve will remain in the diagonal part.

With the emergence of taxes, the expected revenue of producers decreases, and the supply curve will shift from S1 to S2 to the left, and the upper limit of the supply, Gas limit, remains unchanged. At this time, the equilibrium price rises from the original P1 to P2, so the actual fees paid by users not only do not decrease but increase, resulting in a decrease in consumer surplus (the yellow part in the figure). At the same time, the producer surplus of miners (the green part in the figure) also decreases, because the actual fee received by miners is only P*. The blue part is the destroyed base fee. Like the role of taxes, the ultimate beneficiary of this part of the income is all those who hold ETH. Because after the base fee is destroyed, the ETH inflation rate will decrease, and even deflation will occur in the case of network congestion (when the base fee is greater than the block reward), so the price will rise. The red part is called deadweight loss in economics, which refers to the social cost caused by the market not being in the optimal operating state. The direct impact is the decline in the total market production-it could have produced and sold Q1 Gas, but in the end only sold Q2. In other words, some people are unable to complete the transaction because the transaction fee they are willing to pay is lower than the base fee, so they have no way to enjoy the benefits of the transaction. From this, we can judge that in addition to being unable to reduce transaction fees, the 1559 protocol is also unable to solve the problem of network congestion. It only makes the network appear less congested by implementing staggered transactions.

In fact, including the 1559 protocol, any transaction fee mechanism cannot significantly reduce transaction fees. The reason is that transaction fees are determined by market supply and demand. The high transaction fees caused by Ethereum network congestion are essentially a scalability problem, not a transaction fee mechanism problem. If the network throughput cannot be fundamentally expanded, then peak-shifting transactions can only play a temporary role in diverting traffic during peak periods, and the price paid for this is a decrease in transaction volume. Of course, the premise of a decrease in transaction volume is that we do not consider the change in the number of users. In fact, as the number of network participants increases day by day, the number of users will continue to increase in the long run, which will cause the demand curve to shift to the right and further increase the transaction price. Not only that, due to ETH deflation and rising fiat currency prices, the transaction fees actually paid by users may be higher than the theoretical value.

3.2 Whether the volatility of transaction fees is significantly reduced and transaction fees become predictable

Finally, this article discusses a question: Can the 1559 protocol significantly reduce the volatility of transaction fees and make transaction fees predictable? Intuitively, it is feasible, because the base fee can be predicted by the utilization rate of the previous block; although the tip cannot be predicted, with the implementation of staggered transactions, the number of users competing for the same block will decrease, and the total fee should be controllable. In addition, the expected waiting time for packaging will be shortened. It should be noted here that it is the expected waiting time rather than the waiting time. Because of the existence of the base fee, users already know the lowest price before paying. If they cannot accept it, they can just wait for the base fee to fall. It is also waiting, but the difference is that the waiting time under the 1559 protocol is predictable, while the waiting time under the current protocol is endless. It is like a besieged army. If they know that reinforcements will arrive in at most five days, their morale will be boosted; but if they cannot communicate with the reinforcements, they may collapse on the third day. Therefore, the volatility of transaction fees and the expected waiting time for packaging will be greatly reduced.

However, this is only under normal circumstances. Imagine an extreme situation: when the base fee is extremely low, the transaction demand suddenly surges. Since the base fee increase is limited each time, it cannot be significantly increased in a short period of time (for example, within 20 blocks). In order to complete the transaction as quickly as possible, users will pay high tips. At this time, the transaction fee and waiting time become unpredictable. Many studies have proved this. Leonardos, S. et al. (2021) concluded that the increase or decrease of the base fee determines the degree of chaos in the transaction fee. The closer the increase or decrease is to 12.5%, the weaker the dynamic adjustment function of the base fee, and in extreme cases, it will lose the ability to adjust. Moore, IC et al. (2021) found through random process simulation that we cannot reject the hypothesis that the medium- and long-term (more than 4 days) trend of transaction fees belongs to a non-stable process.

In summary, in the absence of extreme events, the 1559 protocol can indeed reduce the volatility of transaction fees; and in the short term, transaction fees will fluctuate according to a certain trend, so they can be roughly predicted, but in the medium and long term, there is no feasibility of prediction.

4 Conclusion

Let’s go back and analyze Eric Conner’s vision for the EIP-1559 protocol point by point:

a) Save up to 90% of transaction costs

In fact, not only has the transaction fee not been reduced, but users may pay higher transaction fees, because although the base fee is a tax on miners, miners will pass part of the tax effect on to users through the invisible hand of the market. Why is it possible? Because users can choose to trade in their spare time to avoid high transaction fees. This will be explained in detail at the end of the article.

b) Improve user experience by automating the transaction fee bidding system

It is feasible because the base fee can be calculated automatically. When the 1559 protocol is officially launched, the wallet will update the corresponding functions.

c) Provide users with a predictable transaction fee system

The short term is predictable, but the medium and long term (more than 4 days) is unpredictable.

d) Reduce the expected waiting time for transaction confirmation

The user's expected waiting time can be reduced, but the actual waiting time has not changed because the transaction volume and throughput have not changed.

e) Even if transaction fees are the main source of rewards, this approach can also discourage selfish mining

It can be achieved. Miners will not artificially create congestion, because they will have to pay a large amount of base fees, and these fees will be destroyed; but miners will not launch empty block attacks due to the destruction of income, because after all, there are still tips to collect. In fact, since the 5.19 crash, on-chain transaction fees have dropped sharply, and the Gas price has remained roughly below 50Gwei. This means that even if the block utilization rate is 100%, the transaction fee reward for a block is only 0.75 ETH, accounting for less than 30% of the total income. Therefore, the miners' currency-based income has not dropped significantly. On the contrary, as ETH appreciates, they can get more income on the gold standard.

f) Allow users to “jump the queue” even when the network is congested

Same as the current mechanism.

g) Consolidate the economic value of ETH at the protocol level

In summary, due to currency deflation, the value of ETH tokens has indeed increased, but the improvement of the ecological value of the ETH network is not significant. The 1559 protocol does not fundamentally solve the problem of network congestion, nor can it reduce the transaction fees of users, and it may even increase them. Of course, users do not need to worry about this. Since short-term transaction fees are predictable, small-capital users can choose to trade in their spare time. At this time, the tax effect is not obvious, and the user's transaction fee expenditure will not be more than under the current mechanism. This is the greatest value of this protocol besides currency deflation; as for large-capital users, they do not need to consider the transaction fee at all. Even at the peak, the transaction fee is just a drop in the bucket compared to the funds they trade. In addition, in addition to currency deflation and short-term predictability of transaction fees, the 1559 protocol has another potential huge function. It is a transit station for the transition of ETH to version 2.0. It gradually moves ETH closer to the 2.0 economic model by continuously reducing the currency-based income of miners and increasing the gold-based income of coin holders.

Finally, let me talk about my subjective judgment. The 1559 protocol is indeed a major change, but its contribution to the price of the currency may not be as great as everyone imagines. After all, value determines price. If Ethereum wants to achieve a qualitative improvement in value, it still has to rely on increasing the throughput of the network, and the impact of supply and demand on price is only a short-term effect. Especially under the current situation, I personally think that it is unlikely to rely on ETH's deflation to restart the bull market. Before 2.0 is truly realized, the bull and bear markets in the cryptocurrency market still depend on BTC.