Dark Forest (Mempool)

What Is the Dark Forest?

The dark forest is a metaphor that describes the hostile environment of public blockchain mempools, where pending transactions are visible to anyone and routinely exploited by automated bots. Just as the dark forest in Liu Cixin's novel depicts a universe where broadcasting your position invites annihilation, the blockchain mempool punishes any transaction that reveals a profit opportunity.

The concept was introduced in August 2020 when Paradigm researchers Dan Robinson and Georgios Konstantopoulos published "Ethereum is a Dark Forest." Robinson had discovered approximately $12,000 worth of Uniswap liquidity tokens accidentally sent to a pair contract. Despite designing an elaborate two-contract obfuscation strategy to recover the funds, a generalized front-running bot detected and replicated the profitable transaction before Robinson could execute it. The takeaway: making a profitable transaction visible in the mempool is effectively handing the profit to predators.

The academic groundwork was laid a year earlier in the "Flash Boys 2.0" paper by Phil Daian and colleagues, which coined the term "miner extractable value" (later renamed maximal extractable value) and documented how bots on decentralized exchanges were replicating high-frequency trading strategies from traditional finance.

How It Works

On most blockchains, when a user submits a transaction, it enters the mempool: a public waiting area where transactions sit until a validator or miner includes them in a block. During this waiting period, the transaction's full contents are visible to every node on the network, including its parameters, gas price, and the value it creates or moves.

Predatory bots continuously scan the mempool, simulating every pending transaction to identify profit opportunities. When a bot finds one, it submits its own transaction with a higher gas fee to ensure it gets included first, or it strategically places transactions around the victim's trade to extract value.

The Predator Taxonomy

Several categories of bots operate in the dark forest, each exploiting different aspects of transaction visibility:

• Front-runners: bots that detect a profitable pending transaction (such as a large DEX trade) and submit a copy with a higher gas fee to execute before the victim, capturing the profit the original transaction would have generated
• Sandwich attackers: bots that place two transactions around a victim's trade, buying before it (pushing the price up) and selling after it (profiting from the inflated price), leaving the victim with worse slippage
• Generalized front-runners: the most sophisticated predators, these bots simulate arbitrary pending transactions, detect any internal profit opportunity, and replicate it by substituting their own addresses, without targeting any specific transaction type
• Liquidation bots: automated systems that monitor DeFi lending protocols for undercollateralized positions and race to submit liquidation transactions, earning the liquidation bonus
• Backrunners: bots that place transactions immediately after a large trade to capture residual arbitrage as prices realign across markets, generally considered less harmful since they do not worsen the original trader's execution

Anatomy of a Sandwich Attack

Sandwich attacks are the most common dark forest predation strategy, accounting for over 50% of MEV transaction volume on Ethereum in 2025. The attack unfolds in three steps:

1. A victim submits a swap on an AMM (for example, buying Token X with ETH)
2. The bot front-runs the victim by buying Token X first, increasing its price on the AMM curve
3. The victim's trade executes at the inflated price, pushing it even higher
4. The bot immediately sells Token X at the new, higher price, pocketing the difference minus gas costs

The most notorious sandwich bot, known as "JaredFromSubway.eth," executed hundreds of thousands of attacks starting in early 2023, with peak daily revenues in the hundreds of thousands of dollars. In June 2026, the bot was itself exploited for $7.5 million through a reverse honeypot attack using fake token contracts.

The Scale of the Problem

The European Securities and Markets Authority (ESMA) estimated in July 2025 that MEV extraction on Ethereum runs at approximately $180 million per month, with Solana adding another $45 million per month through Jito-based auction infrastructure. Post-Merge cumulative MEV on Ethereum has exceeded $1 billion.

Roughly 90% of Ethereum blocks are now built through MEV-Boost, which separates the roles of block builders and proposers under proposer-builder separation. This architecture formalizes MEV extraction rather than eliminating it: specialized builders compete to construct the most profitable blocks, and two to three dominant builders produce 80% to 95% of all MEV-Boost blocks.

Competition among searchers has compressed per-unit profitability over time. The average profit per sandwich attack has fallen to approximately $3, and total monthly MEV profit on Ethereum declined from $10 million in late 2024 to $2.5 million by October 2025. The dark forest has not become less dangerous: it has become more efficient.

Defenses Against the Dark Forest

As MEV extraction has grown, so have the tools to mitigate it. Defenses generally work by hiding transaction contents, redistributing extracted value back to users, or bypassing the public mempool entirely.

Private Transaction Submission

Flashbots Protect is an RPC endpoint that routes transactions through a private channel instead of broadcasting them to the public mempool. Bots cannot see or front-run transactions they never observe. As of October 2024, Flashbots Protect had served 2.1 million unique Ethereum accounts and shielded $43 billion in DEX volume. In December 2024, Flashbots migrated to BuilderNet, a decentralized block-building network jointly operated with Beaverbuild and Nethermind.

MEV-Aware DEX Routing

Some decentralized exchanges and aggregators are designed specifically to neutralize dark forest predation:

• CoW Swap uses batch auctions with uniform price clearing, where all traders receive the same price within a block, eliminating the ordering advantage that enables MEV. Internal analysis showed over a 90% reduction in sandwich and front-running attacks compared to traditional AMM swaps
• 1inch Fusion uses a Dutch auction model where third-party "resolvers" compete to fill orders, processing over $13 billion in cumulative volume since late 2022

MEV-Blocker implements order flow auctions where searchers bid for the right to backrun user transactions, with 90% of the extracted value redistributed to users.

Encrypted Mempools

Encrypted mempools represent a protocol-level defense: transactions are submitted as encrypted ciphertexts and only decrypted after block commitment. Shutter Network uses threshold encryption where decryption requires cooperation of at least a threshold number of independent "keypers." This prevents any single party from reading transaction contents before ordering is finalized. Encrypted mempool proposals are advancing as Ethereum Improvement Proposals in 2026.

Layer 2 and Alternative Architectures

Layer 2 networks and alternative transaction architectures can structurally reduce dark forest risks. Systems that use centralized sequencers process transactions in the order received, eliminating the gas-auction dynamic that enables front-running. Bitcoin-native Layer 2 solutions like Spark avoid the dark forest problem altogether: their transaction models do not expose pending state to a public mempool the way Ethereum does.

The Dark Forest Across Chains

While the dark forest metaphor originated on Ethereum, MEV extraction exists wherever transactions are publicly visible before finalization:

• Ethereum remains the most studied dark forest, with the most mature extraction and defense infrastructure. Approximately 15% of Ethereum transactions now use private RPCs, up from roughly 1% in 2022
• Solana lacks a traditional mempool (validators process transactions continuously), but Jito's client, run by 94% of validators, creates an off-protocol pseudo-mempool where transactions sit for approximately 200 milliseconds, long enough for searchers to bid on front-running and sandwich opportunities
• Bitcoin historically had negligible MEV due to its limited script programmability, but the introduction of Ordinals, BRC-20 tokens, and Runes has created competition for block space that surfaces new extraction opportunities. For a deeper analysis, see the Bitcoin L2 MEV extraction analysis

Why It Matters

The dark forest represents a fundamental challenge for blockchain usability: users lose value on every transaction that interacts with a public mempool. Sandwich attacks function as an invisible tax on DEX traders. Front-running discourages participation in DeFi protocols. And the concentration of MEV extraction among a small number of sophisticated builders raises censorship resistance concerns.

Understanding the dark forest is essential for anyone building or using decentralized applications. Choosing MEV-aware infrastructure (private RPCs, MEV-protected DEX routing, or architectures that avoid public mempools entirely) directly determines how much value users retain versus how much is extracted by bots.

Risks and Considerations

• Private mempools introduce trust assumptions: users must trust the private relay operator not to front-run them or selectively censor transactions, creating a tradeoff between MEV protection and decentralization
• MEV extraction is not entirely harmful: arbitrage bots align prices across markets, and liquidation bots maintain protocol solvency. Eliminating all MEV could reduce market efficiency
• Defense adoption remains uneven: most retail users still submit transactions to public mempools by default, meaning the dark forest disproportionately affects less sophisticated participants
• The arms race continues: as defenses improve, extractors develop more sophisticated strategies, including AI-powered MEV bots that create synthetic wallet networks to avoid detection
• Dark pools and private transaction channels may fragment block space markets, making price discovery less transparent for the broader network

This glossary entry is for informational purposes only and does not constitute financial or investment advice. Always do your own research before using any protocol or technology.