Abstract: This article is a summary of this excellent research paper by Tim Roughgarden. It outlines the connotation of the EIP-1559 proposal, reiterates Roughgarden’s “Ten Key Points”, and tries to explain the arguments in the paper more intuitively.
Background introduction: Ethereum transactions
(If you are familiar with Ethereum, you can skip this part). )
Like all computers, the Ethereum blockchain is a state machine¹. Any given Ethereum state is a “simple” mapping between an address and an account state–the account state is just the data stored in an account (such as Account balance, contract code, etc.), and the account is controlled by a unique address.
Mapping of addresses to their corresponding account states (Source: Ethereum EVM illustrated.)
Transactions are things that change this global state. The transaction specifies a sequence of instructions to change the global state through the execution of the Ethereum Virtual Machine (EVM). (To be more precise, EVM is an implementation form of Ethereum’s state transition function, which is necessary when defining any state machine.)
A very simple state machine. The circle represents the state, and the arrow represents the state transition. Of course, the state machine of Ethereum is much more complicated. Many state transition types are implemented in the form of opcodes, and there may be infinite states¹.
The creator of the transaction must specify the gas limit and gas price.
The upper gas limit is a measure of the cost (computation, storage, etc.) imposed by the transaction on the Ethereum blockchain. The Gas price indicates how much the creator of the transaction is willing to pay per unit of Gas (in ETH). For example, the most basic transaction type (simple transfer) requires 21,000 units of gas; more complex transactions require more gas. Usually the gas price reflects the current demand for EVM calculations and changes by orders of magnitude over time.
(Note: All citations in this article are from the originally cited article
The transaction creator pays: gas limit × gas price.
A block is an ordered sequence of transactions and some related metadata (importantly, it also includes a reference/pointer to the previous block; this is why it is called a blockchain).
The maximum block size is the upper limit of the amount of calculation that any block can contain (represented by the total amount of gas in the transaction). Currently, its upper limit is set to 12.5M gas; therefore, the theoretical upper limit of each block transaction is about 1000 (although in practice it is much less). Miners are responsible for selecting transactions and sorting them into blocks and providing proof of work.
The following are important for understanding the rest of this article:
The transaction fee mechanism (TFM) is part of the agreement, which determines the amount paid by the creator of the confirmed transaction and who gets the fee.
Current transaction fees
The current Ethereum network transaction uses the highest price auction mechanism².
Source: Design of Transaction Cost Mechanism of Ethereum Blockchain-Economic Analysis of EIP-1559
This will result in “many miners packing blocks to the maximum block size, greedily prioritizing pending transactions with the highest gas price”.
EIP-1559: Core Idea
The official Ethereum Improvement Proposal (EIP) #1559 (https://github.com/ethereum/EIPs/blob/master/EIPS/eip-1559.md) can be found here.
Burn the basic expenses related to history
- Each block has a base price (per unit of gas) calculated by the protocol, called the basic fee. Paying the basic fee is a prerequisite for joining the block.
- The basic fee is just a function of the previous block.
- All income from basic fees is burned, that is, it is permanently destroyed from the total circulation of ETH.
Variable block size
- The largest block size is doubled (for example, the upper gas limit is increased from 12.5M to 25M), at this time the old largest block (for example, 12.5M gas) will be used as the target block size.
- As long as the size of the latest block is larger or smaller than the target block size, the basic fee is adjusted upwards or downwards.
- The current transaction is no longer a single gas price, but includes tips and fee caps. Only when the upper limit of the transaction cost reaches the basic cost of the block, the transaction will be packaged into the block.
- Who pays what? If a block with a basic cost of r contains a transaction with a tip of δ, a cost limit of c, and a gas limit of g, the creator of the transaction must pay g·min(r+δ, c) ETH.
- Who received this payment? The income from the basic fees is destroyed, and the remaining part is transferred to the miners of the block.
Interestingly, when reading these key ideas, one might think that they are quite casual, and/or that some of these “key ideas” are a bit orthogonal to other ideas. But in fact, it’s not. Well, not really. These ideas are intrinsically connected, as you will see in the rest of this post.
Interestingly, when reading these core ideas, people may think that they are quite random, and/or that some of these “core ideas” are somewhat contradictory to others. Of course, this is not the case. These ideas are inherently connected, as you will see in the rest of this article.
Variable block size as a proxy for demand
The first point: These ideas are intrinsically linked, and are the concept of dynamically adjusting the block size. Why do you want to do this? There is a simple answer: when the block size is dynamic, the actual size of the mined block can be used as a proxy for demand.
The mechanism of EIP-1559 is to use the past block size as an on-chain measure of demand. Large blocks (more than 12.5M gas) and small blocks (less than 12.5M gas) indicate an increase and decrease in demand, respectively.
Update of formula
The formula for updating the basic cost of each block is recommended as:
In other words, after the largest block (that is, twice the target size), the basic fee increases by up to 12.5%, and after an empty block, it decreases by up to 12.5%. The coefficient of ⅛ is quite arbitrary. A “good” coefficient can adjust the basic cost at an appropriate speed to adapt to the peak of the decline/rise of demand.
Here, I will restate the “ten important points” in Roughgarden’s paper and explain the reasons behind it.
[1.] Any transaction fee mechanism, whether it is EIP-1559 or other mechanisms, is unlikely to significantly reduce the average transaction fee; the continuous high transaction fee is a scalability issue, not a mechanism design issue.
The figure below is a representative example of a supply and demand graph calculated (measured by gas) in the Ethereum network⁴.
Source: Design of Transaction Cost Mechanism of Ethereum Blockchain-Economic Analysis of EIP-1559
Roughgarden believes that all (reasonable) natural gas price mechanisms can be regarded as “working for this ideal.” In other words, try to reach a natural gas price that is closest to the best intersection point. And this point of intersection is entirely determined by the relationship between supply and demand-it has nothing to do with the natural gas price mechanism.
Note that this is completely independent of the gas price mechanism. The intersection of the supply and demand lines is the market clearing price, that is, the price when the total gas demand is equal to the available supply.
Roughgarden believes that all (reasonable) gas price mechanisms can be regarded as “working hard for this ideal”, that is, trying to reach the gas price closest to the best intersection. And this point of intersection is entirely determined by the relationship between supply and demand-it has nothing to do with the gas price mechanism.
Reducing market clearing prices by increasing supply or reducing demand is fundamentally a scalability issue, not a mechanism design issue
[2.] EIP-1559 should reduce the difference in transaction fees and the delay experienced by users through the flexibility of variable block size.
As mentioned in the previous “Points”, as long as demand exceeds supply, transaction costs will be high.
So, what is the significance of this proposal? In order to make transaction costs more predictable, the cost estimation problem, that is, the problem of choosing the best gas price for the transaction, should be as simple and clear as possible.
Essentially, the author believes that the main benefit of EIP-1559 is to improve the user experience (UX) and formalize the user experience through what he calls “User Incentive Compatibility (UIC)”. In order to make this blog post relatively “easy”, I will skip the details of this formalism, but the paper also provides a good intuitive analogy:
Shopping on Amazon is much easier than buying a house in the highly competitive real estate market. On Amazon, you don’t need to pay attention to strategy, and you don’t need to guess yourself; you are either willing to pay the listed price for the products on the shelf, or you are not willing to…
When preparing to buy a house and compete with other potential buyers, you must carefully consider the bid to the seller. Moreover, no matter how smart you are, you may regret your offer afterwards-either because you bid too low and was PKed by someone else, or because you bid too high and the price you paid exceeded your mind. price. The house does not need to be sold to the potential buyer who is willing to pay the most (if the buyer bids too high), this is a loss of economic efficiency.
Sotheby’s art auction. Admittedly, this is not the perfect analogy for the problem we are dealing with: the gas price, but it is indeed a good picture, always an interesting topic in the art world:) Image source: ukartpics/Alamy Stock Photo
In essence, the current Gas price bidding system has caused a lot of “chaos” in the market (and thus inefficient market), because transaction creators usually do not honestly say what they are willing to bid, and their decisions are usually Being influenced by other people (or other people they think). Frankly speaking, setting prices is more direct, thereby reducing market confusion and inefficiency.
[3.] EIP-1559 should improve user experience through simple cost estimates, except during periods of rapid demand growth, in the form of “obviously best bids”.
The part of the paper that proves this is quite complicated, but it essentially boils down to proving the so-called “user incentive compatibility.” The proof in this section (6.3) shows that the 1559 mechanism is generally used as a “price-off mechanism” (as in the Amazon example given earlier). It should be noted that this is not true during periods of rapid demand growth, because:
When the basic fee is too low, users must compete for the scarce block space through their tips, and the 1559 mechanism actually returns to the highest price auction.
In other words, during the sharp rise in demand, 1559 looks like there is no difference in the current system.
[4.] Under EIP-1559, the short-term incentives for miners to execute the agreement as expected are as strong as the highest bid auction.
“As expected” means:
- Miners have no incentive to make fake transactions
- Miners have no motivation to collude with users outside the chain
These arguments are formally proven in Sections 6.2 and 6.4.
[ 5.] Under the EIP-1559 framework, the game-theoretic obstacles to double-spending attacks, censorship attacks, DOS attacks, and long-term revenue maximization strategies (such as manipulating basic costs) seem to be as strong as the highest price auctions.
The explanation for this is quite concise in Section 7.5, but the basic point is: Under EIP-1559, the “main” attack vectors are not easier to exploit than the current gas price mechanism because they are “detectable.” , Is theoretically fragile, or both.”
[6.] EIP-1559 should at least moderately reduce the inflation rate of ETH by burning transaction fees.
Obviously, more burning means less ETH circulation.
[7.] Faced with the threat of off-chain agreements, the two seemingly orthogonal goals of simple cost estimation and cost burning have become inseparable.
Interestingly, the basic cost of burning is a necessary part of the 1559 mechanism, otherwise (that is, if the miners simply charge the fee), the entire 1559 mechanism will be equivalent to the highest price auction; that is, it is now. In other words, if the 1559 mechanism is modified to allow miners to earn basic fees, the current gas price mechanism will be reduced.
I think this is a key point because I have seen other people’s arguments many times that EIP-1559 is the same as the current mechanism, but this is not the case, especially due to the introduction of combustion.
In addition, there is an argument that the basic cost is also necessary in this design, because simply burning the cost from the highest price auction proved to be sub-optimal (because the off-chain agreement becomes profitable)
[8.] Alternative design options include paying basic fee income to miners in future blocks in advance instead of burning them; and replacing user-set variable tips with fixed tips.
This point is also very important. The authors show that all the arguments presented for EIP-1559 in the research paper are also valid for several other alternative designs.
One alternative design that I will specifically mention is to pay the basic cost forward. Due to the proposed fee burning mechanism, the greatest resistance to EIP-1559 comes from miners (see the precautions section below).
[We can prove] In order to make the basic cost of the block economically meaningful, the income from it cannot be passed on to the miners of the block. Perhaps the easiest way to withhold these revenues in the current EIP-1559 specification is to destroy them, effectively issuing a one-time refund to all ETH holders. Another solution is to transfer these revenues to one or more miners in other blocks.
This substitution/modification of EIP-1559 may affect the opinions of miners. I am interested in hearing a wider discussion/debate.
[9.] The basic fee update rules for EIP-1559 are somewhat arbitrary and should be adjusted over time.
Regarding the basic interest rate update function, I have already mentioned it above (“⅛ coefficients are quite arbitrary”). 2 times the maximum block size is also a somewhat random “good” number. More generally, the 1559 mechanism only needs to meet a few conditions, namely:
- Basic expenses related to history, and
- Burn or otherwise withhold the income of block miners
Illustration about overshoot. This picture is from an article on overshoot on Wikipedia. It is intended to be a topic from signal processing, but it is somewhat related to the current topic and can be used to assist in explanation)
In addition, Roughgarden also outlined several additional conditions required for the basic fee update rules:
- After a sudden surge (or decrease) in demand, it is reasonable to quickly adjust upward (or downward).
- The adjustment speed is slow enough to avoid overreacting to small or very short changes in demand.
- A cartel composed of users and/or miners cannot be controlled calmly with game theory.
- The cost of the attacker is high.
How fast is “quite fast”? How expensive is “expensive”? These questions are best answered through experiments and community discussions.
[10.] Variable size blocks make a new (but expensive) attack vector possible: overwhelm the network with a series of largest blocks.
One possibility exists that “aggressive miner cartels” create a series of largest blocks to create demand shocks and then push users to increase tips. This situation is very costly and can also be alleviated by using larger blocks.
However, is there really enough evidence to prove that the harm of miner collusion will not be worse under EIP-1559 than it is now?
It seems that most miners (at least mining pools) have been opposed to EIP-1559. ⁶ This is because they think that burning will lose part of their profits, which is understandable. This brings many questions, the most critical may be: if most miners refuse to include the upgrade of EIP-1559, can this update be implemented? Will this lead to another fork in the network?
Some attack vectors are “fragile games”–for example, keeping the block size at a certain level and artificially reducing the supply–however, as long as there are enough irritated miners and successfully collude, it is possible to implement this This is a “fragile” collusion strategy.
Then someone always asks the question: What impact does this have on the wider community? I don’t think the broader blockchain community will take a positive attitude towards the disputes and forks of protocol changes.
I would like to take this opportunity to re-emphasize the variant of EIP-1559 suggested by Tim Roughgarden: “Prepay the basic cost”. Maybe this can provide a much-needed middle ground?
- EIP-1559 may not cause the average gas price to drop.
- EIP-1559 may improve the user experience (that is, the transaction creator’s experience) through simple cost estimates and less variable gas prices.
- From the perspective of traditional game theory, EIP-1559 provides the same security assurance as the current natural gas price selection mechanism under certain assumptions.
- The caveat in this regard is that greater collusion of miners caused by enough disgruntled miners may undermine some of the more traditional game theory assumptions on which these conclusions are based. Can enough opponents destroy our game theory assumptions? And successfully implemented the otherwise “fragile” collusion strategy?
- Can we have an extensive discussion on the variation of Roughgarden’s EIP-1559: “Prepaid Basic Fees”? Maybe this can provide a much-needed middle ground (between users and miners)?
: To get technical, Ethereum is actually an infinite state machine: https://firstname.lastname@example.org/blockchain-the-infinite-state-machine-ffc39f32e182
: There’s a lot of interesting research about first-priced auctions (and auctions in general, see: auction theory). In particular, a lot of major tech companies that rely on ad revenue have recently switched from second- to first-priced auctions: https://digiday.com/media/buyers-welcome-auction-standardization-as-google-finally-goes-all-in-on-first-price/
: There are some important caveats to this. With the rise of DeFi (Decentralized Finance), there have been increasing cases of miners censoring and reordering transactions in order to profit via arbitrage (and other) opportunities present on DeFi applications. See Section V-B “Threats from Miner Extractable Value” here: https://arxiv.org/pdf/2101.08778.pdf
: The demand curve likely isn’t linear, but it’s not relevant for this argument; i.e. the argument holds for any-shaped demand curve.
: Roughgarden adds: “For example, typical layer-1 scaling solutions like sharding, in which different parts of the blockchain operate in parallel, increase transaction throughput and therefore decrease the market-clearing price. Typical layer-2 scaling solutions like payment channels and rollups, which effectively move some transactions off-chain, decrease demand for EVM computation and likewise decrease the market-clearing price”.
Posted by:CoinYuppie，Reprinted with attribution to:https://coinyuppie.com/summary-of-economic-analysis-of-eip-1559/
Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.