Author: sui14, Compilation: Ladyfinger, BlockBeats
Editor’s Note:
This article provides an in-depth analysis of the impact of the Dencun upgrade on the Ethereum L2 network, revealing positive outcomes such as reduced transaction costs, increased user activity, and asset inflows post-upgrade. However, it also highlights negative effects like network congestion and high rollback rates due to MEV activities. The article calls for community attention and collaborative development of MEV solutions tailored to L2 characteristics to foster a healthy Ethereum ecosystem.
Introduction:
In this article, we aim to provide a data overview of the current state of L2 networks. We monitored the significance of gas fee reductions on L2 due to the Dencun upgrade in March, studied the evolution of activities on these networks, and emphasized emerging challenges driven by MEV activities. Additionally, we discussed potential barriers to developing MEV tools and solutions for L2.
Positive Aspects: Adoption of L2 Post-Dencun Upgrade
Gas costs reduced by 10x after Dencun upgrade
Gas fees on Ethereum L2 consist of costs for executing transactions on L2 and submitting batch transactions to Ethereum L1. Different L2 networks operate with varying gas fee structures and ordering rules based on their developmental stages and design choices. For instance, Arbitrum operates on a first-come, first-served (FCFS) basis where transactions are processed in the order received. In contrast, Optimism (OP Mainnet) and Base, as part of OP Stack, use a priority gas auction (PGA) model combining base and priority fees. Users can opt to pay higher priority fees for faster inclusion and earlier appearance in blocks. Understanding fee structures is crucial for comprehending ecosystem growth and MEV dynamics.
Historically, Ethereum L1 costs constituted the majority of total costs users paid when trading on L2, exceeding 80%, as indicated by the black bars in the graph below. However, after the Dencun upgrade on March 14, L2 shifted from using calldata to a more economical method known as “blobs 1” for batching submissions to L1. These temporary storages include their gas auctions composed of blob base fees and priority fees.
Since Dencun, there has been a significant reduction in fees paid from L2 to L1 — the breakdown of gas costs on the OP Stack chain shows a dramatic change, with L1 costs dropping from 90% to just 1%, while L2 costs now constitute 99% of the total cost. This shift resulted in an overall approximate tenfold decrease in average total gas fees on L2, for example, OP Mainnet’s average gas fee per transaction plummeting from about $0.5 to $0.05.
Surge in Activity on L2
Following cost reductions, there has been a noticeable increase in activity and usage on L2, evident from the skyrocketing L2 gas fees shown in the graph above. Notably, on March 26, Base’s average gas fees exceeded pre-upgrade peak levels. To accommodate more transactions and reduce network congestion, Base increased its gas target starting from March 26 and underwent several adjustments thereafter.
The chart below highlights daily transaction volumes on L2, demonstrating significant growth across networks like Arbitrum, Base, and OP Mainnet. Particularly, Base’s daily transaction volume quadrupled, now processing approximately 2 million transactions daily.
While it’s challenging to determine whether this is a result of organic participation or influenced by incentive programs and Sybil activities — since market conditions improved and the meme coin season sparked by WIF on Solana arrived late last year — all major L2 active addresses and DEX trading volumes have notably increased post-EIP-4844 upgrade, especially on Base and Arbitrum.
Asset Flow to L2
With improved market conditions and the meme coin season sparked by WIF on Solana, total value locked (TVL) on L2 has been steadily increasing since late last year. Notably, Base has emerged as the fastest-growing chain, with recent total TVL surpassing OP Mainnet.
Since early March, Base has seen approximately $1.5 billion USDC flow in, partly due to Coinbase transferring funds from clients and businesses to Base. According to Artemis data since January 2024 across 11 major bridges, there has been a $14 billion outflow from Ethereum to major L2 networks. Arbitrum leads with about $7 billion, followed closely by zkSync, Base, and OP Mainnet. Further data from Debridge Finance, a widely used cross-chain bridge in EVM chains and Solana, confirms Arbitrum and Base as top receivers of all outbound funds.
Negative Aspects: Increasing MEV Activities with Lower Gas Fees
Upon further inspection of transactions, we noticed that bot trading activities are increasing gas fees and rollback rates on L2. In the next section, we’ll conduct a case study using Base’s statistical data to explore this issue comprehensively, highlighting the impact of cheaper gas on L2 after the Dencun upgrade.
L2 Post-Dencun Upgrade: Similar to Ethereum without Flashbots, but lacking transaction pools
Network Congestion
Challenges began to surface on March 26 when daily average gas fees on the Base network briefly surged, surpassing levels seen before the Dencun upgrade. However, by June 3, Base raised its gas target to 7.5M gas/second compared to 2.5M gas/second during Dencun upgrade, effectively reducing average gas costs back to approximately 5 cents.
On the Base network, the most gas-consuming contracts include Telegram trading bots like Sigma and Banana Gun, as well as digital wallets and DEXs such as Bitget and Uniswap. Additionally, many unmarked contracts are involved in token minting, meme coin trading, and atomic arbitrage activities. These contracts rank as top contracts on Base based on gas fee payments.
By comparing the behaviors of popular Telegram bots like BananaGun, it’s clear that transactions executed using these bots incur significantly higher gas fees than regular transactions. After the Dencun upgrade, gas prices spiked to a peak of 30 Gwei for transactions executed using the BananaGun Telegram bot on the Base network. Although this rate subsequently stabilized at around 3 Gwei, it still represented a 43-fold increase in gas costs compared to other transactions.
Daily gas prices on Base, comparison of Banana Gun transactions with other transactions
Analyzing the average gas prices paid monthly by popular DEX trading bots on Base and comparing them with non-Telegram bot transactions (represented in black bars) clearly shows that users using trading bots bear significantly higher gas costs. The comparison of monthly gas prices on Base illustrates the difference between all Telegram bots and other transactions.
Rising Rollback Rates
The rollback rate of transactions in blockchain networks is an important indicator of their health. We noted that after the Dencun upgrade, especially on L2 networks like Base, Arbitrum, and OP Mainnet, rollback rates have increased. Currently, Ethereum mainnet has a rollback rate of approximately 2%, while Binance Smart Chain and Polygon have rollback rates ranging from 5-6%. Before the Dencun upgrade, Base’s rollback rate was also maintained at approximately 2%, but subsequently surged to around 15%, peaking at 30% on April 4. Meanwhile, Arbitrum and OP Mainnet also saw periodic spikes in transaction failure rates, fluctuating between 10% and 20%.
Cross-chain transaction rollback rates
Upon deeper analysis, we found that high rollback rates on L2 networks do not always reflect the actual experience of ordinary users. Instead, these rollbacks are likely caused by MEV bots. By employing the heuristic approach (Query 2), we identified a set of router contracts exhibiting high rollback rates indicative of bot-like behavior when conducting MEV extraction transactions:
Since the Dencun upgrade,
– Active Routers: Contracts processing over 1000 transactions.
– Limited Interaction EOAs: Less than 10 external owned accounts (EOAs) wallets interact as transaction senders.
– Sender Distribution: Less than 50% of transaction senders have only sent one transaction, indicating no long-tail distribution of user groups. This suggests routers are unlikely used by retail users.
– Behavioral Patterns: Transaction history neatly covers 24 hours or shows multiple transactions in one block, indicating non-human behavior.
– Exchange Concentration: Over 75% of successful transactions involve exchanges.
– Detected MEV Transactions: Over 10% of successful transactions use atomic MEV strategies detected by hildobby’s heuristic approach.
Using these criteria, we detected 51 routers on Base, likely representing a conservative lower bound estimate of bot activity on Base.
We divided all transactions processed by routers on the Base network into two groups and conducted comparative analysis. Results showed that router contracts behaving like bots exhibited significantly higher rollback rates compared to other transactions: bot-like contracts averaged a 60% rollback rate, six times higher than the approximately 10% observed for other transactions.
Daily rollback rates on Base, comparison of bot-like contracts with other transactions
Based on the data above, we can infer that automated trading activities like MEV bots and Telegram bots are likely major contributors to high gas costs and rollback rates on the Base network.
The single sequencer architecture of L2, combined with the absence of public transaction pools, fosters extensive MEV strategies utilizing sequencers, which become the primary cause of network congestion. This congestion is particularly pronounced in L2 networks employing priority gas auction (PGA) mechanisms like OP Mainnet and Base. The result is not just network congestion but also substantial block space and gas costs wasted due to rolled-back transactions and MEV searcher activities. This parallels the emergenceThe previous situation on Ethereum was similar; however, unlike Ethereum L1, sandwich MEV phenomenon does not exist on L2 due to the current absence of transaction pools.
What is the scale of MEV on L2?
Understanding MEV activities on L2 networks is crucial for assessing their impact. Currently, there is no widely accepted figure for L2 MEV, validated through multiple sources and reliable methods. Additionally, unlike Ethereum L1, L2 lacks real-time monitoring tools like mev-inspect, libmev, and eigenphi, which are essential for measuring total MEV and miner profits.
Some of the released L2 MEV datasets and studies include:
– An open-source dataset constructed by hildobby on Dune Analytics (inspired link: Sandwich | Sandwich | Atomic Arbitrage)
– The research paper “Quantifying MEV On Layer 2 Networks” by Arthur Bagourd and Luca Georges Francois, quantifying MEV on Polygon, OP Mainnet, and Arbitrum using mev-inspect and funded by Flashbots.
– The research paper “Rolling in the Shadows: Analyzing the Extraction of MEV Across Layer-2 Rollups” by Christof Ferreira Torres, Albin Mamuti, Ben Weintraub, Cristina Nita-Rotaru, and Shweta Shinde, quantifying activities and discussing new MEV strategies on L2 including sequencer roles and batch confirmation delays.
In addition to these resources, Sorella Labs will soon release their MEV data indexer tool Brontes, an open-source repository available for Ethereum L1 and L2. Flashbots and the Uniswap Foundation are seeking funding to expand L2 MEV taxonomy and quantification. For those working in or interested in collaboration, please contact the Flashbots Market Research team.
Although further validation is needed, the dataset published by hildobby on Dune Analytics provides a valuable initial benchmark.
Volume of atomic arbitrage on L2 using hildobby’s dataset
Over the past year, atomic arbitrage MEV trading volumes on six major L2 networks including Arbitrum, OP Mainnet, Base, Zora, Scroll, and zkSync exceeded $36 billion, accounting for 1% to 6% of decentralized exchange (DEX) trading volumes on each chain. Initially concentrated on Arbitrum and OP Mainnet, these MEV trading volumes have recently shifted towards Base and zkSync.
Compared to atomic arbitrage trading volumes, Sandwich attack volumes on L2 networks are significantly lower, contrasting sharply with Ethereum where Sandwich attack trading volumes are four times that of atomic arbitrage. This difference is primarily due to L2 networks’ use of a single sequencer setup without transaction pools, limiting the ability of sandwich MEV exploitation unless there is data leakage from transaction pools or sandwich attacks initiated by a single sequencer. Thus, on L2, strategies such as atomic arbitrage, flashbots, statistical arbitrage, and liquidation have become more viable for searchers.
Breakdown of Ethereum MEV volume
How to measure the remaining MEV income in the MEV market on L2?
While quantifying the MEV market precisely is challenging, we can compare numbers from other ecosystems with MEV solutions for size comparisons:
– On Ethereum L1, annual validator income from MEV-boost blocks is approximately $96.8 million (based on an estimated price of $3,500/ETH); the median value of MEV-boost blocks is four times that of regular validator blocks.
Distribution of block rewards between regular and MEV-boost blocks
– On Solana, validators collect additional MEV income through Jito’s bundling service from validator tips, estimated at approximately $338 million annually (based on a price of $130/SOL).
Daily tips earned through Jito’s bundling service, by validators and Jito Labs
While the exact total MEV volume on the Base network has not been disclosed, estimating the market size can be done by observing the income of the Banana Gun Telegram Bot, one of the most active participants in the market. Banana Gun’s trading volumes on Base L2 network and Solana are roughly equivalent, generating over $1 million in daily trading volume on each chain, equating to over $10,000 in transaction fees per chain per day.
Banana Gun Telegram Bot, cross-chain volume and fees
Note that Banana Gun Bot’s market share on Solana may differ significantly from Base. For example, there are several other major Telegram Bots on the Solana platform, such as Sol Trading Bot and BonkBot, while Base may support fewer Telegram Bots. Therefore, one cannot directly use the proportion of Banana Gun’s trading volume and MEV income on Solana to estimate the total MEV income on Base.
However, another predictive approach shows different results: in March, Banana Gun Telegram Bot paid over $23 million to Ethereum’s block builders and validators. Specifically, during the week of March 26 to April 1, trading volumes on Base actually exceeded those on Ethereum, as indicated by the peak in the chart, suggesting significant potential for MEV income on the Base network. This comparison of cross-chain trading volumes reveals Base’s growth prospects in terms of MEV.
Certainly, there are significant differences between Base and Ethereum in the MEV ecosystem. MEV competition on Base may be less intense than on Ethereum, potentially resulting in lower fees paid by Bots bidding to validators. Nevertheless, meme coin trading Bots relying primarily on sniper and arbitrage mechanisms remain viable under Base’s sequencer architecture.
MEV income paid by Banana Gun Telegram Bot users to validators
Focus on MEV strategies and solutions in L2 networks
Ethereum has developed a mature MEV ecosystem equipped with infrastructure tools serving participants at various levels of the supply chain. At the protocol level, MEV-boost allows validators to outsource block building tasks through auction. For searchers, bundling services offered by Ethereum block builders — akin to Solana’s Jito Labs and Polygon’s FastLanes — enable them to implement MEV strategies including rollback protection. These services ensure that block builders simulate transactions and only execute those that are determined not to roll back. Additionally, private RPC services like Flashbots Protect provide ordinary users with ways to bypass public transaction pools and their potential risks. However, current L2 networks have significant room for improvement in developing MEV infrastructure comparable to this.
Why focus on MEV strategies and solutions in L2 networks?
The MEV phenomenon persists in environments lacking transaction pools and plays a crucial role in maintaining market efficiency, especially through strategies such as statistical arbitrage, atomic arbitrage, and liquidation in outdated AMM and lending markets. However, the lack of mature MEV infrastructure such as bundling services may lead to some negative consequences. Without transaction pools, many MEV strategies may devolve into spam strategies, leading to increased network rollback rates and exacerbated network congestion.
By implementing bundling services, focusing MEV competition from the main chain to auxiliary chains can effectively alleviate the burden of high gas fees faced by users due to MEV robot competition. Meanwhile, searchers benefit from rollback protection, enjoying higher profit margins and reducing the risk of failure costs.
For L2 networks using shared sequencers, mainstream solutions often require users to submit transactions to public transaction pools, which may lead to the recurrence of sandwich attacks. In such cases, MEV protection tools like Flashbots Protect are particularly important, as they not only protect users from the threat of sandwich attacks but also may provide refunds for MEV or priority fees, ensuring users receive higher-quality transaction executions and more favorable prices.
The development of complex MEV infrastructure faces some unresolved challenges. Firstly, as more value flows to sequencers, searchers’ profit models change over time, and marginal profits may decrease. This change may raise questions about the sustainability of long-term highly competitive search strategies. We expect market mechanisms to adjust to this phenomenon, where common search strategies pay a larger but not complete proportion of value to sequencers, while less common strategies pay less.
Additionally, existing MEV infrastructure, such as Ethereum’s block building market, is evolving rapidly in terms of order flow dynamics. So far, these factors have been major drivers of the centralization trend in the block building market and the rise of private transaction pools on Ethereum L1. Ensuring the competitiveness and fairness of the block building market remains an issue that needs to be addressed.
Finally, MEV solutions on L2 networks may need to be distinct from current Ethereum mechanisms, primarily due to L2’s unique features: such as shorter block generation times, lower block space costs, and relatively centralized governance structures. For example, Arbitrum’s block time is only 250 milliseconds, and whether such rapid block generation rates can be compatible with existing MEV infrastructure remains uncertain. At the same time, ample and economical block space provided by L2 has greatly changed the pattern of transaction search, making spam problems more serious and urgently requiring new solution strategies. Furthermore, compared to other environments like Ethereum L1, L2 governance is more centralized, which may allow additional requirements to be imposed on MEV service providers, such as requiring block builders to avoid sandwich attacks on users to ensure market fairness.
