The Cost of Network Resilience

Authors: Elaine Hu, Alejo Salles, and Hasut
Compiled by: Qianwen, ChianCatcher

Key Takeaways

• The new mev-boost feature will allow validators to locally build low MEV blocks to maximize Ethereum's censorship resistance while still outsourcing the construction of high MEV blocks.
• Utilizing this feature comes with an opportunity cost, which is the price of gaining resilience.
• Since most blocks are low EV, sacrificing a bit of profit greatly aids in enhancing resilience.
• This is a breakthrough that allows one to avoid the trade-off between maximizing MEV extraction and maximizing censorship resistance.

Network Resilience

After the Office of Foreign Assets Control (OFAC) expanded its list of sanctioned addresses to include smart contracts, Flashbots recently became a target of public scrutiny, affecting many users of the Ethereum network. While Flashbots has tried to navigate the risks and ambiguities posed by OFAC's sanctions on data and communication infrastructure in web3 from the beginning, this issue was not as concerning before the merge. Prior to the merge, relayers sent bundles to mining pools, which assembled complete blocks locally. The market gradually shifted towards full block submissions, completely separating block builders and validators, who could no longer add individual transactions to blocks.

Currently, most Ethereum blocks comply with OFAC regulations, although in practice, this means that only a small fraction of non-compliant transactions can be curtailed. Ensuring strong decentralization and resilience at the protocol level is crucial for Ethereum's success as a neutral infrastructure. However, merely changing mev-boost is insufficient to provide these assurances. For consensus clients to profitably convert between blocks from builders (which may be selected) and the local mempool, both execution and consensus clients need deep reforms.

While the community agrees this is an important issue, client teams have different priority options to weigh. Other long-term proposals, such as Inclusion Lists and partial block auctions, are part of the core "Proposer-Builder Separation" research.

So what can we do today to address this issue? Launch a robust, protocol-level solution simultaneously. Some have suggested that Flashbots self-limit its share of relayed blocks, but we believe such market-distorting behavior would yield little result, as we cannot truly predict the behavior of other relayers.

Others have proposed that Flashbots charge a small fee to encourage other entities to run independent relayers. While this seems like a wise suggestion, it could potentially entrench the role of relayers, so we view it only as a temporary solution until SUAVE is deployed—a fully trustless building network.

Fortunately, other builders and relayers have begun to emerge and are capturing an increasing share of the market, and we are actively supporting teams interested in running other relayers. Since last week, we have opened our builder code, allowing anyone to easily run a builder now.

However, from the validators' perspective, the issue remains: should they run mev-boost and comply with the regulations adopted by their trusted relayers, or should they only build blocks locally, which could mean sacrificing over 100 ETH in profit opportunities?

Solutions

Now, validators can run mev-boost with the new min-bid parameter to sidestep this issue. This parameter will only accept blocks from relayers if the bid for the block is above a selected value; otherwise, it proposes a locally built block. For most blocks, mev-boost does not yield much profit. Therefore, validators can sacrifice a small amount of profit in exchange for a more resilient network while remaining open to high MEV opportunities.

As described, the new min-bid parameter allows validators to avoid the trade-off between profit maximization and network resilience. In the following sections, we will explore the costs associated with validators choosing this feature.

Block Profit Distribution

First, let's look at the cumulative distribution of block profits, including blocks from Flashbots relayers and locally built blocks, with all data being post-merge:

As expected, mev-boost exhibits a long-tail distribution. We can illustrate the profit differences between the two types of blocks, viewing this data from a new perspective, as shown below:

This asymmetry in distribution again shows us that mev-boost blocks are generally more valuable than mempool blocks. However, we also see that a considerable number of blocks have almost no difference in value. For example, with a profit scale of 0.05 ETH, historically, 35% of mev-boost blocks had profits of 0.05 ETH or less. Setting the minimum bid value to 0.05 ETH means that proposers will choose to build local blocks about one-third of the time.

The question now is, if validators choose to set the min-bid parameter to a positive number, how much profit do they need to sacrifice?

Opportunity Cost

The opportunity cost of setting a minimum bid refers to the cumulative profit difference between blocks using the Flashbots mev-boost relayer and locally built blocks. The higher the minimum bid, the more local blocks are built, and the greater the opportunity cost.

As we see from the graph, with a high minimum bid parameter, the opportunity cost becomes quite significant, meaning most blocks will be built locally. However, as we pointed out, we are primarily focused on low-profit blocks, which can help us go further in achieving network resilience. We focus only on blocks with profits below 0.11 ETH, which account for 73% of all mev-boost blocks. As shown in the following graph, we can understand the opportunity cost and the approximate annual interest rate implied by each parameter setting:

We can see that the opportunity cost increases with the minimum bid, but at the same time, the predicted annual interest rate decreases very slowly. This raises the question of whether sacrificing a small amount of profit will impact transaction inclusion, which we will examine next.

Impact on Transaction Inclusion

Currently, about 72% of Ethereum blocks comply with OFAC standards, meaning some transactions experience delays when being included on-chain. In particular, blocks containing these transactions have about a 28% probability of being included after one block, corresponding to validators who build local blocks or profit from non-compliant relayers. If all validators use mev-boost and only connect to compliant relayers, the inclusion rate would drop to 0.

Let's look at the inclusion rates of these transactions under different min-bid values, focusing only on mev-boost instances connected to compliant relayers:

As mentioned earlier, the default zero value prevents any transactions from being included. However, for relatively small min-bid values, we see that even after very few blocks, non-compliant transactions have a significant probability of being included. For example, setting the minimum bid to 0.05 ETH increases the probability of a block being included in another block from 0 to 35%. Clearly, if most validators choose to use this feature, it will greatly enhance the network's resilience.

The following table summarizes the different outcomes for various values of the minimum bid parameter:

What to Do?

Although the issue of censorship ultimately needs to be resolved at the protocol level, we can leverage the fact that most blocks are low profit today, meaning the profit difference between locally built blocks and relayer-built blocks is not significant.

If you run an Ethereum validator and are willing to accept a small opportunity cost in exchange for making the network more resilient, please run mev-boost with -min-bid \<x> -relay.

Where x is the minimum acceptable profit from relayer blocks, measured in ETH. As we saw in the table in the previous section, setting x = 0.05 results in an opportunity cost of 0.011 ETH per block, with the proposer's annual yield only decreasing by 0.2%. This slight decrease in yield has a significant impact on network resilience: if all validators connected to compliant relayers adopt this threshold, the inclusion rate of non-compliant transactions will increase from the current 28% to about 53%. A greater willingness to sacrifice profit will lead to greater network resilience.

We encourage validators to update mev-boost and help Flashbots achieve its ongoing commitment—to build an open and decentralized infrastructure that aligns with Ethereum's goals.