Ethereum’s Censorship Controversy
| By Sachin Lal, Kyle Maulden, and Morgan Pare
TLDR
- Credible neutrality is an essential component of the Ethereum network, but there is growing concern within the community that this neutrality is threatened by MEV-Boost and OFAC regulations.
- By comparing mempool data with on-chain data, we found that potential OFAC censorship has no impact on what transactions make it on-chain.
- OFAC censorship does have an impact on the time for a transaction to be added to the blockchain, with OFAC non-conforming transactions taking marginally longer than their OFAC conforming counterparts.
- While some transactions experienced slight delays, the impact on user experience is minimal and the concerns over censorship are overblown.
Credible Neutrality
A fundamental aspect of Ethereum is its emphasis on operating as a credibly neutral network. In a blockchain network, credibility is of utmost importance as the network’s value is derived from the trust users have in its operations. A credibly neutral network is one that is impartial, fair, and transparent, where no one entity has control over the network’s operations. This ensures that the network operates according to a set of predefined rules and is not susceptible to manipulation by any individual or group. By being credibly neutral, Ethereum can attract developers and users who seek a trustworthy platform to build and deploy their dapps and smart contracts. This helps to create a thriving ecosystem of decentralized applications, which in turn increases the value of the network. As more people begin to use Ethereum for transactions, the network’s credibility becomes even more critical to maintaining trust in the system.
As Ethereum transitioned to PoS with the Merge, Flashbots introduced MEV-Boost in June 2022, an open-source software, to democratize the extraction of MEV and make it accessible to all validators. Validators who run MEV-Boost are connected to a network of Searchers, Builders, and Relays who specialize in identifying, extracting, and transferring MEV to Validators. Upon introduction of MEV-Boost, Flashbots launched the first relay which filters blocks based on OFAC regulations.
What is OFAC
OFAC, or the Office of Foreign Assets Control, is a financial intelligence and enforcement agency of the US Department of the Treasury. It is responsible for implementing and enforcing economic and trade sanctions based on US foreign policy and national security goals against targeted foreign countries, terrorists, international narcotics traffickers, and those involved in activities related to the proliferation of weapons of mass destruction. Businesses and individuals must ensure that they do not facilitate transactions with Specially Designated Nationals (SDN) or blocked persons, as it could be a violation of OFAC regulations and result in significant fines and penalties.
It is crucial for validators seeking to adhere to US regulations to have a good understanding of OFAC regulations. Although OFAC regulations do not provide a clear guideline for compliance, some validators may choose to exercise caution and refrain from proposing blocks that contain transactions involving individuals or entities listed on the SDN list.
Censorship Concerns
With over 90% of validators adopting MEV-Boost, relays that filter blocks based on the SDN list (i.e. OFAC filtering relays) are a significant concern for Ethereum’s credibly neutral network. The ability to censor transactions based on observance of US sanctions raises questions about who gets to decide what is or isn’t a legitimate transaction. This could lead to a scenario where certain transactions might remain pending for extended periods or may never be included in a block, which would directly conflict with the network’s goal of being open and accessible to all. Overall, the introduction of Flashbots’ MEV-Boost and OFAC filtering relays sparked concerns within the Ethereum community about the network’s credibility and neutrality.
Flashbots and the community have taken several steps to improve the censorship resistance of the network. For example, Flashbots open sourced their relay and introduced a min-bid setting enabling validators to only accept MEV-Boost blocks above a certain threshold. Additionally, the community has stepped up, and currently seven out of eleven relays in operation do not filter at all based on OFAC regulations. Despite all these attempts the community is still troubled and concerned with censorship of Ethereum as many cite MEV Watch, which claims over 55% of blocks enforce OFAC filtering.
Considering all the controversy, we decided to assess the effect of these OFAC filtering relays on Ethereum’s credible neutrality.
Our Approach & Results
To understand the prevalence of SDN transactions and the current impact of OFAC sanctions on the neutrality of the network, we analyzed mempool and on-chain transactions for a one week time frame between 2023–02–09 06:32:00 UTC to 2023–02–16 06:32:00 UTC. We did not consider internal transactions for this analysis.
Think of the mempool as a holding space for transactions waiting to be included in a block and added to the blockchain. It provides insight into all transactions submitted to the network before any potential censorship takes place. By comparing transactions in the mempool with those seen on-chain, we can gauge the extent of transaction censorship.
It is important to note that there is no singular, global mempool as each node operates its own. Discrepancies between the transactions of the mempools of different nodes arise from variation in network latency or node configuration differences. For this analysis, we sourced the mempool transaction data from a single node on Ethereum mainnet as we believe this gives a representative sample.
When transactions enter the mempool, they can be in one of two states: pending or queued. Pending transactions are ready for processing and are eligible for inclusion in the next block. Queued transactions, on the other hand, are not ready for processing as they are out of sequence. Our analysis solely consisted of pending transactions as they are in a valid state to be included on-chain and thus exit the mempool. Furthermore, we removed all canceled transactions from our analysis, by filtering out all transactions that had duplicate transaction nonce values from each sender.
We defined a transaction to be non-conforming with OFAC regulations if the sender or receiver address was listed on the SDN list. In reality, relays and builders go a level deeper than this by looking at the internal call stack, but this is a good proxy. We refer to these as OFAC Non-Conforming Transactions. Conversely, we define OFAC Conforming Transactions as transactions where both the sender and receiver address is not on the SDN list. The following table summarizes some of our findings:
We saw a total of 7,494,287 distinct pending transactions in the mempool, 848 of which, or ~0.01% were OFAC non-conforming. Nearly all (99.52%) of these OFAC non-conforming transactions were later seen on-chain, while only 95.32% of OFAC conforming transactions were later seen on-chain. Therefore, the SDN list does not appear to have a significant impact on which transactions end up on-chain.
Our next step was to analyze the duration of time that these transactions typically remained in the mempool to determine if OFAC non-conforming transactions took longer to be added to the blockchain. We found that OFAC non-conforming transactions spent a median of 22 seconds in the mempool. In contrast, OFAC conforming transactions had a median of 8 seconds in the mempool. The figure below highlights the contrast in mempool time between the two types of transactions.
As seen from the plot above, OFAC conforming transactions typically exit the mempool and make it on-chain more quickly than OFAC non-conforming transactions.
One possible explanation for this phenomenon could be MEV-Boost, which has been embraced by more than 90% of validators and provides access to a block building marketplace. Many of the top relays, such as Flashbots or Blocknative, filter based on OFAC regulations. Thus, if a block builder adds OFAC non-conforming transactions to a block, they would have to resort to less popular, non-conforming relays, which lack sufficient validator registrations. This might cause OFAC non-conforming transactions to remain in the mempool for longer periods and face delays in being added to the blockchain.
However, there are also a number of other confounding effects that could contribute to the difference in time spent in mempool and warranted further exploration. First, and most obvious are the fees of these transactions. With higher fees, a transaction is more likely to be included on-chain in subsequent blocks, and therefore exit the mempool more quickly. Consequently, we compare the time spent in mempool against the max fee per gas and the priority fee per gas of successful type 2 transactions (a transaction type proposed in EIP1559) from our sample. In each figure, the shaded regions represent the 25th to 75th percentiles of time spent in mempool, while the solid lines represent the median value.
As expected, the time in mempool generally declines as the max fee and priority fee increase for both OFAC conforming and OFAC non-conforming transactions. In addition, OFAC conforming transactions consistently have a lower time spent in the mempool than their OFAC non-conforming counterparts. This holds true across all the entire range of the max fee per gas and the max priority fee per gas. Furthermore, the set of OFAC conforming and set of OFAC non-conforming transactions in our sample have relatively similar distributions across these fee ranges. Therefore, the differences in time spent in mempool cannot be attributed to differences in fees.
Secondly, we found that type 2 transactions are more likely to be included in the upcoming blocks than type 0 transactions (legacy transaction format prior to EIP1559). In fact, in our sample, 85.4% of type 2 transactions were included in the next block; for type 0 transactions, this number is only 65.5%. A further comparison of time in mempool for successful transactions by transaction type is shown in the figure below.
It is clear that for both OFAC conforming and non-conforming transactions, type 2 transactions are included on-chain more quickly. We hypothesize this is because of the increased control on a transaction’s fee structure that type 2 transactions provide.
In our sample, OFAC conforming transactions were type 2 83.1% of the time, while OFAC non-conforming transactions were type 2 only 69.4% of the time. The higher percentage of type 2 transactions means that more OFAC conforming transactions were likely to be included on-chain promptly, simply due to their transaction type. Therefore, the transaction type is likely a contributing factor in the difference in time spent in mempool seen for OFAC conforming vs. OFAC non-conforming transactions.
Conclusion
Through the course of our analysis, we showed that a transaction’s status as OFAC conforming or non-conforming does not have a material impact on the transaction’s likelihood of making it on-chain. However, it appears that OFAC non-conforming transactions spend more time in the mempool in a pending state resulting in a longer time to make it on-chain. That being said, the difference is not substantial. Based on the median time spent in mempool, OFAC conforming transactions are typically included in the next block, whereas OFAC non-conforming transactions are typically included in the block after that. This disparity is likely due to minor network censorship caused by MEV-Boost and other factors including transaction type.
Censorship resistance remains a critical issue for our community and we must remain vigilant. However, a closer look at the data revealed that while some transactions experienced delays, the impact on user experience was minimal. This suggests that some worries have been exaggerated and unfounded. Nonetheless, we recognize the need for ongoing monitoring and have decided to track metrics that expose potential censorship attempts in real-time. Please join our Discord and keep an eye out for additional tools as we unveil them. This will enable us to enhance the continued integrity, neutrality, and security of Ethereum.
