Section III: Flashbots: Taming the Dark Forest

These user-facing protections emerged partly because the industry recognized that individual defenses alone weren't enough. By 2020, Ethereum faced exactly this market chaos at scale. The priority gas auctions described earlier were creating network congestion, while miners were capturing MEV through opaque, off-chain deals that favored well-capitalized participants.

Enter Flashbots, a research organization founded in 2020 with a radical proposition: instead of trying to eliminate MEV, create transparent infrastructure to make it more fair and efficient. Their insight was that the current system was wasteful, and that channeling extraction through better infrastructure could reduce harm.

MEV-Geth and the First Solution

In January 2021, Flashbots released MEV-Geth, a modified Ethereum client that let miners accept transaction bundles over a private Flashbots channel instead of only from the public mempool. Rather than spamming ever-higher gas bids in priority gas auctions, searchers could submit bundles directly to miners running MEV-Geth. Miners simulated and ranked these bundles and included the most profitable ones in their blocks. This moved most of the competition off-chain, cutting down on bidding wars in the public mempool while still letting professional searchers compete for MEV opportunities.

The Transition to Proof-of-Stake

When Ethereum moved to proof-of-stake in September 2022 (a transition detailed in Chapter II), the entire MEV landscape needed rebuilding. Flashbots developed MEV-Boost, an open-source middleware that provides Proposer-Builder Separation (PBS), a design where specialized builders construct blocks and validators simply choose which block to propose, rather than validators doing both jobs themselves. This expanded the builder-validator relationship introduced earlier into a full competitive marketplace via relays. As of early 2026, approximately 90% of Ethereum blocks are built via MEV-Boost.

This separation currently exists outside Ethereum's core protocol, implemented through MEV-Boost rather than built into the blockchain itself. Researchers continue working on enshrined PBS, which would make proposer-builder separation a native part of Ethereum, but that work remains in development.

How MEV-Boost Works

This process is facilitated by trusted entities called relays. Relays act as a neutral escrow and auctioneer: builders send them full blocks, and the relay verifies their validity and bid. The relay then forwards only the block header and the bid to the proposer (validators are also called proposers in this context). The proposer chooses a header without seeing the block’s contents, preventing them from stealing the MEV opportunity. The system evolved from individual miners making direct deals to a sophisticated auction where multiple builders compete for validator selection, with relays facilitating the bidding process.

These trust assumptions are not just theoretical. In April 2023, a validator exploited a vulnerability in MEV-Boost and relay handling to “unbundle” private bundles, copy profitable MEV transactions, and siphon more than $20 million from other searchers in a single block. The episode triggered urgent client patches and became the basis for the first high-profile U.S. criminal case about MEV infrastructure: in 2024, federal prosecutors charged two MIT-educated brothers with wire fraud and money laundering for allegedly orchestrating the exploit, with the case still being litigated as of early 2026. Whatever one thinks of the legal theory, it underscored that MEV relays and builders are no longer just technical plumbing but also legal and regulatory attack surfaces.

Expanding User Protection

Recognizing that infrastructure alone was not enough, Flashbots also launched the user-facing protection service mentioned earlier: Flashbots Protect. By routing transactions through private channels that bypass the public mempool, ordinary users gain protection from sandwich and frontrunning attacks while potentially receiving rebates from captured MEV. These transactions still compete in the builder auction but are never exposed to public mempool predation.

The Flashbots approach represents a pragmatic philosophy: given that extraction is baked into how ordering markets function, the goal should be making it transparent, efficient, and less harmful. Rather than fighting the economic forces, they built infrastructure to channel them constructively. However, this infrastructure-based solution revealed an uncomfortable truth: organizing MEV markets efficiently also created powerful chokepoints that concentrated control in unexpected ways.