All you need to know about EIP-1559 (and more)

Table of contents

Language options: Japanese

What is EIP-1559?

EIP-1559 is primarily a large change to how users buy transactions on Ethereum (though it has already been adopted by other blockchains such as Filecoin, Celo, and NEAR). 

Right now, users have to submit a bid (denominated in gwei/gas) to the network. When a block is found, miners fill up the block with the top n transactions, and each transaction pays what they have bid. 

This is called a first-price auction, and the correct strategy for participants is complex and error-prone. This leads to frequent under- or overpayment, as well as increased mental transaction costs.

EIP-1559 gets rid of the first-price auction and largely replaces it with a different mechanism, called a fixed-price sale. The protocol will quote a price, and the user can either pay that price or not, like they would buy an item on Amazon.

A second large benefit derives from the fact that quoted price (called basefee) is no longer paid to miners but is burned instead. All public blockchains need to attract miners (in PoW) or stakers (in PoS) for their security. Today, how much Ethereum spends on its security is strongly linked to the congestion of the network.

When congestion is high, fees are high, and hence the miner incentive is large. But when congestion goes up, that doesn’t mean demand for security goes up as much, and then Ethereum would spend more on security than necessary. This comes at the cost of ETH holders, who subsidize the security via dilution from the newly minted block subsidy of 2 ETH per block + uncle rewards.

After EIP-1559, holders would continue to pay the block subsidy when congestion is low. But when congestion is high, they don’t need to pay for security because the demand of transactors already pays for it. So they get an implicit refund via the burning mechanism.

There are some other minor benefits. If you want to learn more, our analysis of EIP-1559 is a good start. After that, we strongly recommend Prof. Tim Roughgarden’s analysis.

^ Table of contents

Why do you care?

I’m a researcher with a focus on the economics and security of public blockchains, primarily Bitcoin and Ethereum. EIP-1559 affects both, and is one of the largest and most exciting upgrades inside my area of expertise. 

The more I studied the mechanism and its possible alternatives (usually with Georgios Konstantopoulos, Research Partner at Paradigm), the more I became sold on the idea. I’m convinced that EIP-1559 is great for every blockchain that adopts it, and have recently become more involved with Ethereum governance to see it through. 

^ Table of contents

Is EIP-1559 good for ETH?

There are several reasons to believe that EIP-1559 should be good for the price of ETH. 

First, the improved transaction fee mechanism improves the overall experience of using Ethereum and could lead to higher adoption at the margin.

Second, there is the straightforward argument that burning the basefee internalizes some of the revenue (in ETH) that miners currently make. This turns ETH from a non-productive asset into a productive asset, like a company starting a new business line that produces a lot of revenue.

Third, by lowering ETH’s inflation rate and possibly even making it deflationary, ETH becomes more attractive as a storehold of wealth asset and hence could vastly expand its target market. This narrative is already being picked up by mainstream financial media here or here.

These arguments are generally underappreciated, especially by the mining community. We aren‘t just playing in a zero-sum game between miners and ETH holders for distribution of fees, but Ethereum itself is in competition with dozens of other protocols for what we think is a gigantic target market. Growing the pie benefits everyone.

^ Table of contents

Will EIP-1559 lower gas fees?

Fees are primarily the result of supply and demand, not how the marketplace works. The structure of the marketplace can have some minor effect, e.g. reduce overbidding. But it could also have the opposite effect – due to the better UX, users could be willing to pay more for transactions than they are today.

Ultimately, the total capacity in a block remains the same, and the supply of transactions doesn’t change. Only more scalability can lower gas fees in the long-term.

^ Table of contents

How does EIP-1559 impact miner revenue?

With EIP-1559, what we currently know as transaction fees will be split into two parts: The basefee (set by the protocol) and the tip (set by the market). The basefee of each transaction will be burned, while the tip will continue to go to the miners. 

Image: What happens to current miner revenue sources post EIP-1559

The impact this has on miner revenue isn‘t as easy to predict as you would initially think, for two reasons:

First, we don‘t know what the split between basefee and tips will be. Until recently, most people thought (and still think) that almost all fees would be burned, but this won‘t be the case. 

That is because gas does not equal gas. The existence of markets on top of Ethereum (such as exchanges, lending protocols, etc.) creates a constant stream of financial arbitrage opportunities. These opportunities, e.g. to arbitrage the price between two Defi exchanges or between one Defi and one Cefi exchange, to liquidate borrowers or margin traders, and so on, have tremendous financial value. But they can only be performed by the first trader who gets them mined on the blockchain.

That is why the first 10-20 transactions in every block tend to pay disproportionate amounts of transaction fees – much more than is necessary to get included in the block. They do this because they must be early in the block to complete their trade. Miners currently benefit tremendously from these transactions, many of them without even knowing about it. 

This revenue would be unaffected in EIP-1559, because it would be paid via the tip, not the basefee, and hence isn’t burned. Georgios and I have recently quantified the importance of these “priority transactions” relative to normal transactions, and made a surprising discovery.

Depending on how much MEV is correctly classified by our data source (MEV-Explore v0, powered by Flashbots), we can see that miners are likely making more money from selling priority gas today than from selling regular gas. As a result, less than half of today’s fees could be burned by EIP-1559.

Image: Subsidy vs fees vs MEV depending on how much MEV we correctly classified

Second, miner revenue also depends heavily on the price of ETH. For example, if EIP-1559 lowers miner revenues (in ETH) by 30%, and the price of ETHUSD increases by 43%, miner revenue would be exactly the same in USD.

There are good arguments why EIP-1559 should be indeed very positive for the price and adoption of ETH, see Is EIP-1559 good for ETH?

^ Table of contents

As a miner, how do I benefit from MEV?

A popular question I’ve seen especially from smaller miners is whether they individually benefit from the MEV that Georgios and I showed miners are making today, or if they need to take further action to get it.

The name MEV (miner extractable value) is misleading in that sense since it implies miners need to actively do something. But the answer is that no further action is necessary. The financial arbitrage opportunities that result from Defi and tokens on top of Ethereum lead to priority gas auctions in the public mempool and the fees from these gas auctions accrue to all miners equally.

To give you a concrete example, here is a recent block that paid 112 ETH in rewards. 2 ETH from the subsidy, the rest from fees. On further inspection, we can see that over 70 ETH of that actually comes from only seven transactions, at index 1, 2, 5, 6, 7, 10, and 12 respectively. These transactions are primarily Compound liquidations.

^ Table of contents

How does EIP-1559 impact Ethereum security?

At least denominated in ETH (though not necessarily in USD), Ethereum will pay miners less money than before. We expect this to lower the hashpower pointed at Ethereum. However, this doesn’t meaningfully decrease its security.

The reason is that above a certain threshold, each unit of security has decreasing marginal utility. At first, the network and its users benefit tremendously from every miner. But the more miners are already there, the less benefit every additional miner makes.

Image: Money paid to miners has a decreasing marginal benefit.

Ethereum has a permanent block subsidy of 2 ETH per block + uncle rewards. The subsidy is set in a way that it alone should always be sufficient to keep the network secure. 

Image: The block subsidy creates a baseline of miner revenue that should be enough to keep Ethereum secure.

Ethereum has a social contract called Minimum Viable Issuance. The logic behind it is that Ethereum needs a security budget that scales with the network’s demand for security. If ETH doubles in price, so should the amount it pays to miners. So one motivation is to keep Ethereum secure, but the second (conflicting) motivation is to pay as much for security as necessary, but no more.

We can show that this motivation has to exist, using a simple thought experiment. A network whose native token inflates 100% year over year has a very large security budget. But the high inflation is a tax on holders, making the asset unattractive to hold. When a blockchain’s native asset is unattractive to hold, it has a very difficult time bootstrapping itself in the marketplace.

Image: Underpaying and overpaying for security are both bad.

Due to the rise of congestion and the large fees, Ethereum has found itself in the “overpaying for security” part of the graph. Many people (especially miners) have pointed out that Ethereum will lose hashpower due to this change. They may be right about that (the real answer is, it depends), but wrong to see this is an undesirable side effect. There is such a thing as having too much hashpower, considering that someone has to pay for it. And so losing some hashpower is indeed the goal.

This argument isn’t isolated to Ethereum itself. When other networks keep their security at an adequate level and burn the overlay, ETH will be less attractive in comparison.

In conclusion, while lowering miner revenue also lowers security at the margin, Ethereum doesn’t just optimize for security but also usefulness. A system that overspends on security is less useful to its holders.

Hence the equilibrium is for all networks to burn fees that are not necessary for security.

^ Table of contents

Why does basefee have to be burned?

There are some reasons we have to burn the fees for the mechanism to work, and why we want to burn fees for other reasons. We’ll start with why we want to burn fees.

First, burning the basefee gives us the opportunity to reduce Ethereum’s linkage between congestion and security budget. Holders (and the market at large) can tolerate a larger permanent block subsidy when they know that if transactional demand is high, the subsidy will be partially or entirely reimbursed via burning. 

Second, several studies [1] [2] have shown that when miner revenue is dominated by variable transaction fees instead of a predictable block subsidy, this can cause instability of the blockchain itself.

That said, the burning is also required for the EIP-1559 mechanism to work in the first place. To prove this point, imagine that EIP-1559 wouldn’t burn the basefee but simply pay it to miners. The price quoted by the protocol is no longer binding for miners and transactors. If the protocol quotes a price of 100 gwei, and users want to pay 80 gwei, and there is still room in the block (which there almost always is in EIP-1559), then users could pay 100 gwei and then the miner refunds the 20 gwei in a side-channel.

Some have suggested that this problem can be navigated by paying the basefee not to the miner who included the transaction, but instead collecting it in a fund that then pays the rewards forward in time to other miners (like a secondary block reward.)

While this seemingly solves the problem of refunds assuming uncoordinated miners, it also creates an incentive for miners to form a single entity, or several large entities, so they can benefit from the said refunding strategy. It also doesn’t solve the first two problems.

^ Table of contents

Can miners block the deployment of EIP-1559?

Even though miner revenue has increased over 10x in the last year, individual miners haven’t necessarily become more profitable, since the hashrate and subsequently difficulty have also increased. This shows you can never pay too much money to individual miners, but you can pay too much money to miners as a whole.

Like every public blockchain, Ethereum pays miners an incentive to attract a sufficient amount of hashpower (or stake, in PoS) to secure the network. But as we discussed here, Ethereum indeed overpays for security at the expense of ETH holders, who subsidize security via the block subsidy. So it is entirely rational (and positive for Ethereum as a whole) to change that.

In spite of these arguments, some miners are threatening to oppose EIP-1559 based on the ground that it could lower their revenue when denominated in ETH. How much does their opinion matter in practice?

To look at this question, Georgios and I performed a scenario analysis that found the best outcome for rational miners is to simply accept EIP-1559. There are still at least 18 months of PoW mining left, and so miners have too much future revenue at stake to jeopardize it with an attack that could undermine the demand for ETH and to transact on Ethereum.

Finally, it is not even clear if miners will make less money after EIP-1559 has been deployed. Accounting for MEV, we found that as little as 20% of current miner revenue could be burned. And that number doesn’t even account for a potential ETH price increase as a result of EIP-1559.

The bottom line is the same as with all mining-related questions in public blockchains: In theory, miners could hurt the protocol via double-spend or censorship attacks. But in practice, miners can make more money by cooperating with users than by undermining them. We have little doubt that the same incentives for cooperation are in place with regard to EIP-1559.

^ Table of contents

How big are the UX benefits for users?

The main difference EIP-1559 will make for transactors is that they no longer have to guess a transaction fee that is high enough for swift inclusion but low enough so it doesn‘t overpay. Instead, the protocol will quote a price, and the user can either pay that price or not, like they would buy an item on Amazon. They can also submit a lower fee and wait for the price to go down in the future.

The reason this dynamic is possible is that instead of fixing the quantity of gas per block, Ethereum blocks will start to have a lot of extra room (100% extra room, to be precise), and so anyone who wants to transact at the current price usually can. Some or even all of that extra room can be used if people are willing to pay, but it will cause the basefee to go up very quickly until the fee is again high enough that transactions no longer fill more than half the block.

For more on this, and our best guess for how often the system will degrade to the current system of first-price auctions, see my thread here.

^ Table of contents

Does EIP-1559 accelerate MEV extraction? 

Some people have pointed out that EIP-1559 could accelerate MEV extraction due to the following dynamics:

  1. Most miners are not actively extracting MEV today, although they gain most of the same benefit from the fees of priority gas auctions. The discussions around EIP-1559 put MEV on their radar for the first time.
  2. Some miners feel entitled to their current revenue, even though it is the result of users overpaying for security. As a result, they might respond to the deployment of EIP-1559 with user-harming behavior that was previously seen as taboo.

First of all, we have to distinguish between two forms of MEV: benign and malicious. 

Malicious MEV is one that causes chain instability, censors users or key infrastructure (e.g. preventing oracle updates), or even directly steals user assets (e.g. censoring a rollup fraud-proof). 

Benign MEV is pretty much everything else, ranging from financial arbitrage between exchanges, liquidations, frontrunning, etc.

One important insight is that benign MEV can be extracted from non-mining parties, and this is happening today on a large scale. Only miners can extract malicious forms of MEV, and this is generally not happening today, and shouldn’t happen in the future. 

However, if miners started to extract the benign MEV that is currently extracted by non-miners, that would be largely positive for Ethereum. There are three arguments for that:

First, non-mining arbitrageurs are generally a tax on the system – they pull money out, without giving anything back. If miners stood to receive the same arbitrage profits, they would compete for them with higher hashpower, thereby making Ethereum more secure. So the MEV would directly pay for security, which it does only to a lesser degree today via the priority gas auctions. 

For users, it doesn’t make a difference: Someone will extract the benign MEV from them anyway. So naturally, they would prefer if the money goes to miners, where it also secures the system.

In response to this, one could argue that normalizing MEV extraction by miners instead of non-miners creates a slippery slope where miners don’t just extract benign MEV, but ultimately also malicious MEV. This would greatly hurt Ethereum, and we continue to need strong social norms against it.

The second argument why Ethereum benefits from miners extracting benign MEV is that MEV extraction by non-miners creates unnecessary waste on the Ethereum blockchain, which could be used for non-MEV use cases. Flashbots has found that at least 4-5% of all Ethereum blockspace is currently used by MEV transactions, including many failed transactions. As usual, Flashbots only establishes a lower bound – the real number is probably a fair bit higher.

The third argument is that the most sophisticated miners are already extracting MEV directly, either by including their own arbitrage transactions or by selling blockspace privately (not publicly) in return for private order flow. 

This is not good for Ethereum because it turns Proof-of-work into Proof-of-MEVextraction – a much more complex game where some miners can have large benefits over others. The miners better at extracting MEV will grow at the expense of the miners who are worse at it, which presents a serious centralization vector in mining (and staking later on, so this problem doesn’t go away with the move to PoS either.)

The solution to this is indeed education and transparency into MEV, and to normalize the extraction of benign MEV by miners to level the playing field. EIP-1559 does accelerate that, but it’s a trend that needed to happen anyway.

For more information on the topic, read A Collective Effort in Bringing Transparency to MEV by Flashbots.

^ Table of contents