In-depth analysis of the Ethereum block space market: New patterns will emerge in the mining cycle

Written by: Leo Zhang, founder of Anicca Research and Georgios Konstantopoulos, research partner of Paradigm. The author authorized Lianwen to publish the Chinese version.

“A freely functioning market is like a freely rotating wheel: it requires an axle and well-lubricated bearings. How to provide that axle and ensure that those bearings do not starve for lubrication is the problem that market design solves.”
―Alvin E. Roth, The Sharing Economy: Market Design and Applications

Blockspace Market Overview

Blockspace is the commodity that drives the heart of all cryptocurrency networks. In the blockspace market, miners are producers, mining pools are auction houses, and users are bidders. The impact of the blockspace market is so pervasive that it touches nearly every aspect of the cryptocurrency ecosystem.

Whenever a user initiates a transaction, it is broadcast peer-to-peer in the memory pool of each node. Each transaction is accompanied by a fee. The fee represents the willingness to purchase block space so that it can be processed and included in a block.

At any given moment, there are many "proposed blocks" in this "Schrödinger state", between unconfirmed and confirmed, competing to find the first hash output that meets the difficulty target. Each block has the possibility of becoming the next block. By contributing billions of calculations per second, miners gain the upper hand in the battle against the probability wave and earn rewards from filling in the ledger history.

Since the block size is capped, a limited number of transactions can pass through at a given time, giving block space an implicit time value. Transactions that go unconfirmed for too long may be affected by market fluctuations or be preempted by arbitrage bots. The fees users pay to purchase block space reflect their willingness to bid for their space and time. The block space market is where miners and users connect.

On the surface, the blockspace market appears complex and chaotic due to the lack of central coordination. It relies on detailed rules, procedures, and the convergence of supply and demand to self-regulate. How do we know: Is the current market design optimized to ensure success?

Nobel Prize winner in economics Alvin E. Roth is considered a pioneer in the field of market design. In his seminal book, Who Gets What - and Why, he argues that for markets to work properly, they need to do at least three things:

1. Market depth: There are enough potential buyers and sellers to interact. In the block space market, block rewards incentivize suppliers to provide hash rate. On the other hand, as more people use the network to trade, the demand for block space increases.

2. Security: Market participants must feel comfortable disclosing or adopting confidential information they may hold. Due to the transparent nature of on-chain transactions, users who submit sealed bids may not always receive the results they desire. In addition, transactions require a high degree of settlement assurance. That is, there should be enough hash rate to make it expensive to reorganize historical blocks.

3. No congestion: Transactions should be approved or canceled in a timely manner. If the market cannot effectively cope with the congestion caused by transaction volume, participants may not be able to include their transactions in the block without too much delay. As can be seen from the recent popularity of Ethereum DApps, Gas prices have also skyrocketed due to network congestion.

Throughout the history of Bitcoin and Ethereum, the design of the optimal blockspace market structure has often sparked heated debates. In the following, we will demonstrate the structure of the Ethereum blockspace market from the perspective of the supply side (miners) and the demand side (users). We review whether the current blockspace market design can provide depth, relieve congestion, or allow safe and simple participation. Next, we discuss popular proposals for optimizing market structure and how the blockspace market will develop in the future.

Supply side: Ethereum mining structure

The ultimate purpose of the entire mining industry is to act as a decentralized, transparent clearinghouse for a single commodity: block space.

This task is by no means easy. In a distributed system that operates 24/7 around the world, without an authoritative coordinator, miners need to ensure network security by investing a huge amount of hardware and electricity costs to generate an incredible amount of computation to provide strong settlement guarantees for anonymously auctioned block space.

Despite widespread concern about the concentration of hashrate, the mining industry is not a concerted effort. Price fluctuations in the block space market affect each of these components to varying degrees. Each miner is heavily influenced by factors such as geographic location, machine type, temperature, machine maintenance, and mining strategy.

Bitcoin and Ethereum mining have very different market dynamics. Bitcoin and many proof-of-work tokens have almost completely migrated to the ASIC mining era. Ethereum, on the other hand, has almost no ASIC mining despite being the second largest token by market cap. Although there are frequent rumors of new ASIC miner manufacturers launching Ethereum miners, it is estimated that currently around 80-90% of the hash rate in the Ethereum mining industry is dominated by GPUs.

Structurally, how does a market of block space providers that is primarily comprised of GPUs differ from one that is primarily comprised of ASIC miners?

Every GPU mining story starts with NVIDIA and AMD. These two companies either sell discrete graphics cards directly or wholesale GPU chips and memory to downstream graphics card manufacturers. These manufacturers remove features that are not related to mining, so these are called "mining graphics cards". There are many white-label custom mining graphics card brands on the market.

There are significantly more custom GPU miners on the market than ASIC miners. As a result, Ethereum’s hashrate composition is more diverse than a typical ASIC network. This also poses a challenge to distributors monopolizing the channel. Miners have more options because they don’t have to wait for ASIC miners such as Bitmain and Shenma to solve supply chain problems. This means that initial supply is usually not as often subject to supply bottlenecks as ASIC miners using advanced back-end processes.

Furthermore, GPU miners are not tied to any particular network. Miners engaged in speculative mining almost exclusively use GPU mining rigs. In addition, retail GPU cards are also valuable in other computing areas, such as gaming, data centers, and AI work. Overall, GPU miners' hardware has a higher option value.

Structure determines properties. The hardware composition determines the industry’s average capital expenditure and energy consumption. These two factors are critical to calculating mining costs and have ripple effects on the rest of the mining ecosystem: from manufacturing, distribution, hosting facilities, gas cost fluctuations, preferences for EIP proposals, to the defining characteristics of its mining market cycle.

Supply side: Ethereum mining market cycle

In The Alchemy of Hashpower, we introduced the four basic stages of the mining market cycle based on the relative rate of change between Bitcoin price and network hash rate:

Source: The Alchemy of Hashpower, Pt II.

Due to the inherent illiquidity of the hardware market, changes in network hash rate tend to lag behind price changes. "Hardware reaction time" depends on various external factors, such as manufacturers' supply chain bottlenecks, wafer foundry inventory, factory capacity, and even transportation logistics.

These delays are particularly important this year as the market is running at full speed into a "bull market". Miners and investors are scrambling to order new machines. On the other hand, manufacturers are just beginning to recover from supply chain disruptions during the COVID-19 pandemic, and the global shortage of integrated circuits is forcing all semiconductor businesses: mining, automobiles, consumer electronics, etc. to line up in long queues for wafer allocations.

In addition, NVIDIA recently announced that they will artificially destroy the Ethash algorithm mining performance on the latest graphics cards to prevent miners from buying up all GPU inventory. This means that unless the token price or fees start to soar astronomically from now on, the two stages of "inventory-flush" and "The Shakeout" may become a nightmare for many miners when the backlog of mining machines finally come online.

In the Ethereum blockchain, mining revenue comes from three main sources:

1. Coinbase Rewards (2 ETH + Uncle Block Rewards for each block mined)

2. Transaction fees, and

3. Miner Extractable Value (MEV) (more on this later)

Ethereum's transaction fees as a percentage of total block rewards is much higher than Bitcoin's. This means that Ethereum miners not only pay attention to the token price, but also to the price trend of Gas. Even if the ETH price remains stable, the increase in transaction fees is enough to motivate miners to increase hash rate.

Data source: Coin Metrics

In addition, as mentioned in the previous sections, due to the flexibility of GPU option value and distribution, it is easier to scale up or down the hash rate compared to ASIC networks. Therefore, Ethereum mining market cycles tend to be shorter:

Data source: Coin Metrics

Shorter cycles mean that competition accelerates quickly when mining revenue is high, while the option value of hardware means that hashrate is easier to split when mining revenue is low. Unit profitability (block rewards per MH/s) can fluctuate wildly, making it difficult to predict earnings:

Data source: Coin Metrics

Significant increases in token prices and transaction fees attract more miners to join the mining. However, unlike most commodity markets, more producers do not mean an increase in block space supply. The supply of block space depends on the block size and the average block time. This means that an increase in hash rate will not reduce network transaction fees, but it will increase the network's security budget. As more miners join the competition, it becomes more expensive to reorganize historical blocks, thereby improving the network's security guarantees.

Demand side: Time value of block space

Since the block size is capped, participation in a permissionless manner requires competition for system resources. The fee paradigm is the most critical core user experience for anyone initiating an on-chain transaction.

Ethereum, the most popular storage and execution smart contract platform, has experienced rapid development in terms of its practicality. DeFi "money Lego blocks" have spawned products and services that can interlock without permission and have greatly promoted the innovation of new financial mechanisms.

Today, Ethereum users participate in the block space market through repeated first-price auctions. This is a simple auction format where users submit bids to have their transactions included in the next block, which are paid to miners in the form of transaction fees. Users can choose their bids by the "gas price" of the transaction (expressed in gwei / Gas [1gwei = 1e-9 ETH]).

Empirical observation of mining pools’ transaction selection methods shows that more than 75% of mining pools follow the default strategy with no prioritization. That is, transactions are simply included in descending order of fees without prioritizing any specific addresses:

Source: Ethereum Gas Price Statistics

The market structure is simple: users want to minimize the fees they pay to miners in order to enjoy a smooth experience, while miners want to maximize their revenue since they are for-profit entities.

It goes without saying that the fees paid by users will always follow the law of supply and demand: the block space produced per second is a scarce asset, so users who want the next block to include their transactions immediately will always pay more fees than users who are willing to wait. This can be seen from the shape of the pending transaction queue in the figure below:

Pending transaction queue, source: GasNow

Blockspace is the closest thing to "digital real estate" in the cryptocurrency space. Blockspace has the intrinsic value of "real estate" and economic activity occurs within it.
For miners, the time value of future block space is low due to the uncertainty of token price, network difficulty and fees. For users, the time value of future block space is low due to the uncertainty of its profitability and transaction utility.

Demand side: congestion and costs

The time value of block space directly translates into the amount of fees paid by users. In this fee paradigm, estimating the “correct” gas price is a difficult problem, as is evident from the fluctuations in gas prices over the time period of a single block.

Over the course of a week, the median gas price fluctuated from 100 Gwei to 400 Gwei.

Source: corelab

Recall the three elements of successful market design mentioned by Roth: market depth, security, and non-congestion. Obviously, when transactions are congested, the gas price usually rises too high for ordinary users. Where is the bottleneck?

This unpredictability stems from the fact that users cannot negotiate the right fee to be included in the next 1, 5, or 10 blocks. Today most users bid in a "one-shot" manner: they broadcast a transaction once and then wait for it to be included in a block by a miner. By allowing users to express their fee preferences within a certain range, improvements to the constant factor can be achieved, such as using a fee upgrade algorithm.

The initial release of a new technology is always rough. Over the years, various workarounds have emerged to address Ethereum’s congestion issues. Market design for block space involves balancing the interests of many factions in the ecosystem. Currently, three possible paths have been discussed:

1. Short term: Block size increase. A stopgap measure that may compromise security.

2. Medium term: Change the auction mechanism. Community consensus is required.

3. Long term. Scalability solutions. Rollups and ETH 2.0.

One might be tempted to increase the size of a single block so that it can accommodate more transactions (i.e. increase supply assuming demand remains constant). This change would only temporarily ease the pain of high fees, as new demand would quickly fill up, increasing fees again. Additionally, increasing block size makes blockchain node software more resource intensive, so this approach should be avoided if the decentralized nature of the system is to be maintained.

Another approach is to restructure the bidding process. Today, the fee mechanism in all blockchains implements a "universal first-price auction". Ethereum's EIP-1559 proposal changes the mechanism to a fixed-price sale when the system is not congested, making it easy for users to choose the "best" bid for their inclusion preferences (as opposed to the status quo). Variants of EIP-1559 have been deployed in Near, Celo, and Filecoin. The EIP-1559 proposal is scheduled to be activated in Ethereum in July 2021, which will be the largest fee market mechanism change ever for a public chain.

EIP-1559 is also one of the most controversial topics in Ethereum to date. With more than 60% of mining pools expressing opposition to the proposal, EIP-1559 has turned into a "trade war" between users and block space producers. Although it is difficult to quantify the exact impact on miner fee income at this time, the mining industry generally believes that Gas prices need to rise significantly to make up the difference. Ajian, who criticized the proposal in the Chinese Ethereum community, believes that the EIP-1559 proposal "will lead to a loss of miner loyalty." Although Poolin has not yet announced any official position, its founder believes that this will not actually have much impact on mining revenue, but "it is extremely insulting."

However, not all miners feel the same way. The founder of F2Pool, an active user of DeFi products, agrees with the proposal. Even Jihan Wu, a central figure in the Bitcoin block expansion war, expressed support for the change. Ultimately, human coordination remains the biggest challenge in an unstructured open ecosystem rooted in all parts of the world.

In the long run, the right solution is to allow the supply layer to scale horizontally without substantially affecting the trust requirements of the L1 public chain system. After all, users want low fees, and low fees are not an economic mechanism design problem, but a fundamental problem of blockchain scalability. It is worth noting that scalability will allow more demand to enter the system, which will offset the benefits of reducing the average fee per transaction. The main approach here is the so-called Layer 2 L2 solutions, such as Optimistic and ZK Rollups.

Dark Forest and Dark Pool

We are discussing this kind of thing in Bitcoin, and the discussion around the demand side ends here. But an entire financial system is being built in Ethereum that is a game-changer. Ethereum may have certain "opportunities" for a limited time, specifically arbitrage activities waiting to be arbitrageured, or limited participation in asset sales (such as ICOs) where demand is high. Similar to people competing to line up to buy limited edition clothing, developers have built software that can intelligently compete for on-chain business opportunities, even competing with other bots by setting gas price bids.

This concept is called the Priority Gas Auction (PGA), which was first proposed and described in the seminal paper Flash Boys 2.0 by Phil Daian et al. The "irregular" income stream generated by the PGA is called "Miner Extractable Value" or "MEV". Bots participating in the PGA have access to very lucrative opportunities, and they are willing to bid high gas prices to obtain the highest profits.

In Ethereum is a Dark Forest, the author describes a robot MEV extraction operation in which approximately $12,000 of tokens fell into the hands of "predators." These predators are arbitrage bots that constantly monitor activity in the memory pool and try to front-run specific types of transactions based on predetermined algorithms. DEX platforms such as Uniswap may be rife with arbitrage bots.

As a result, some service providers have begun offering “dark pools” — transactions that bypass the public memory pool and are therefore invisible to the public until they are written to the chain. These dark pools do not broadcast to the network, but instead relay transactions directly to miners so that they are not broadcast to other nodes on the network. These dark pools are not entirely for profit maximization purposes. In the article “This time we fought our way out of the Ethereum DeFi dark forest and perfectly saved $9.6 million”, security researcher samczsun documented how his team used this technique to save 25k ETH from a vulnerable smart contract.

Front running and dark pools are not unique to the cryptocurrency market. They represent a financial driver as old as time: secrecy. Wall Street has long been at odds with this controversial beast. What appears to be a dense market to a human, with hundreds of trading opportunities in a single second, looks sparse to a computer. It is estimated that after 2015, dark pools accounted for 15-18% of trading activity in U.S. exchange-listed securities.

The “total potential market” of MEV has grown exponentially, with at least $350 million in MEV extracted since the beginning of 2020, a third of which occurred in February 2021. Most of the extracted MEV is mainly distributed in arbitrage activities in popular AMMs such as Uniswap, Sushiswap, Curve, and Balancer, while a small part is attributed to liquidations in Compound and Aave.

Cumulative MEV extracted since January 1, 2020. Source: Flashbots

The butterfly effect is amazing: arbitrage and liquidation opportunities create MEV. MEV competes through PGA. Fee estimators use the gas price pushed up by PGA as a reference, causing users to overpay transaction fees for their own transactions. Fundamentally, the problem stems from the fact that users and bots are in the same transaction pool, regardless of whether they are chasing MEV.

Ideally, MEV transactions should be in a separate transaction pool from non-MEV transactions. This way, bots that extract MEV can compete with each other, while all other transactions, such as CryptoPunk transfers, will be conducted in a pool without MEV, which will make the Gas market less volatile.

Unfortunately, such a huge change to Ethereum would be difficult to execute. A simpler way to solve this problem would be to introduce a new API endpoint for miners that would allow them to accept bundles of MEV-only transactions. This way, traders could submit transactions directly to this endpoint, while users could continue to use the rest of the system as they do today. This is the approach taken by Flashbots, and as far as we can tell, it causes the least disruption.

How do miners deal with MEV?

In the past, when fees were a negligible percentage of the total block reward, the main focus of miners was to capture as much block share as possible. Miners would choose a hosting pool with a sufficiently large hashrate. After several years of development, the mining pool infrastructure has been optimized to more or less the same performance range. Pools that lag behind their competitors can be easily identified. Miners don't really care about being picky about pools, except for basic parameters (luck, payment schedule, and pool fees). Users don't really care which pool takes their transactions, and mining pools don't care who the users are as long as the fees are attractive.

As MEV grows relative to block rewards, the considerations of miners, mining pools, and users become more nuanced. With MEV, the block space market mechanism will shift from a pure commodity market to one that incorporates some elements of a matching market. When users submit transactions, they should be concerned about the mining pool's ability to execute the transaction in a timely manner.

All other things being equal, MEV causes Gas prices to be higher than they would be without MEV. This means that miners are already indirectly extracting a small portion of all MEV taken by hunting bots, currently estimated to be around 12%.

A recent analysis also suggests that fees earned from MEV will eventually exceed fees earned from "regular" revenue streams (if they haven't already!). Miners are for-profit entities and may want to capture more MEV. Miners may choose to specialize in internal transactions. Using the considerable power they have over transaction ordering, they may choose to insert transactions, reorder transactions, or even audit transactions to maximize the profitability of their transaction operations.

In theory, miners will continually attempt to reorganize the blockchain as they attempt to extract MEV from historical blocks (commonly known as a "time bandit" attack). While this is possible, it is clearly not feasible: miners are structurally long ETH (in most cases), and such an action would directly negatively impact their ETH investment. This can be summarized by saying that any theoretical miner attack will not be executed if it is detected by users.

A more optimistic view is that miners decide to outsource MEV mining to traders. They can do this via the Flashbots approach mentioned earlier: retaining the block rewards and outsourcing only the sorting, or by renting their hashrate to specialized traders.

Regardless of the approach, we expect that Ethereum mining pools will inevitably begin to become more active in the MEV extraction process, seeking to earn the best yield for their miners’ hashrate. As competition for MEV extraction heats up, it remains to be seen how this change in market structure will affect the (de)centralization of the Ethereum mining ecosystem.

How can other members of the ecosystem participate in MEV extraction?

The financialization of hash rate is an important underlying trend in the mining space. If block space is analogous to real estate, then hash rate is the equity in the property, and a forward contract for hash rate is analogous to a mortgage. For miners, selling their hash rate through hashrate tools is a way for them to lock in future income. Similar to renting hash rate on a cloud mining platform, forward contracts allow miners to sell a fixed amount of hash rate for a certain price over a period of time.

Active DeFi users who expect network MEV activity to accumulate can speculate by purchasing these hashrate tools. For example, a trader who expects network MEV activity to surge in the next 15 days can purchase a certain amount of hashrate at a prepaid price in the next 15 days. During the contract period, mining rewards and MEV income earned by miners will then be transferred to the trader. If the trader bets right, the transaction may earn back the MEV fees that the trader paid to the mining pool himself. Both hashrate tools with liquidity and MEV represent the most cutting-edge developments in the mining capital market. The market is actively building new tools to move in these directions. MEV plus hashrate tools realize a complete closed-loop capital market cycle for users and miners.

Summarize

Ethereum blockspace market structure is an interesting topic to study, and for good reason. In this article, we observed the key differences between the supply side of GPU and ASIC hashrate markets. We also identified the key mechanisms that create demand-side momentum. We then bundled the two together under the umbrella of miner extractable value.

As Charlie Noyes wrote in MEV and ME, “Any attempt to prevent miners from capturing revenue streams is likely to incentivize the creation of extra-protocol markets.” MEV is the inevitable result of the increasing complexity of Ethereum transaction types and the current design of the block space market. As the tools and knowledge base for MEV become more mainstream, more interesting MEV patterns will emerge. These behaviors will profoundly affect how often DeFi users plan their block space purchase strategies.

Changing the way goods are sold in the market changes how goods are produced. MEV opens up new avenues for mining revenue, which in turn will affect how the mining industry interacts with block space buyers. New patterns will emerge in the Ethereum mining market cycle.

Thanks to Tina Zhen, Tarun Chitra, Hasu, and Alex Obadia for their insights and comments on an earlier draft of this article.