Skip to main content

This site is for educational purposes only. Nothing here constitutes financial advice.

Lesson 2 of 8
~22 min↑ On-Chain Research

Lesson 2 — The mempool and pending state: visibility, MEV, and private submission

Once you've broadcast a transaction, it sits in public view before being included. Today: what's in the mempool, who watches it, and when to bypass it.

Intermediate
Evergreen
22 min readUpdated 2026-05-18Block Clarity Hub Editorial Team

Every transaction starts in the mempool — the queue of broadcast-but-not-yet-included transactions visible to anyone running a node. Most users never think about this stage. MEV operators think about little else. Understanding what's visible, to whom, and what it means for your transaction-execution quality is the difference between blindly broadcasting and informed submission.

What the mempool actually is. Every Ethereum node maintains its own view of pending transactions. When you broadcast a transaction through MetaMask, your RPC provider (Infura, Alchemy, or a private endpoint) submits it to the network, where it propagates to other nodes. Each node holds these transactions in memory while validators / builders select which to include in the next block. The mempool is *public* in the sense that anyone running a full node can see every pending transaction — which means anyone with bot infrastructure can analyse and react to your transaction before it's included.

Why the public mempool creates problems. Three categories of MEV (Maximal Extractable Value) operate on the mempool. Front-running: a bot sees your pending DEX swap, submits an identical transaction with higher gas fees, and gets included first — capturing the price impact you would have caused. Sandwich attacks: a bot front-runs you with a buy that pushes price against you, then back-runs you with a sell, capturing the slippage. Generalised arbitrage: bots watch the mempool for any transaction that creates an arbitrage opportunity (a swap, a liquidation, a mint) and race to capture it in the same or next block.

Private mempools. Several systems let you submit transactions privately — they go directly to block builders without appearing in the public mempool first. Flashbots Protect is the most-used: your transaction goes to Flashbots' bundle stream and is included in a block without ever being publicly visible. MEV-Share is a more nuanced variant that allows transaction senders to share information about their trades with searchers in exchange for a portion of the MEV captured — turning the front-running problem into a negotiated revenue split. MEV-Blocker and CoW Swap offer similar private-flow options on top of solver-based execution. For large DEX swaps (anything above a few thousand dollars), private submission is the difference between paying slippage to a bot and not paying it.

What you can see in the mempool yourself. Tools like Blocknative's mempool explorer, Eden's mempool dashboard, and mevscan.io let you watch the public mempool in real time. You can search for pending transactions from specific addresses, see gas-price distributions, and observe MEV bots reacting to opportunities within milliseconds. For research purposes, watching the mempool during a high-volatility event (a token launch, a large liquidation, a protocol exploit in progress) is extremely educational — you see the bot ecosystem operate at full speed.

Pending vs confirmed transactions. A transaction in the mempool is not on the blockchain yet. It can be: (a) included in a block — becoming a confirmed transaction; (b) dropped from the mempool — usually because its gas price was too low and it sat too long; (c) replaced by the sender — submitting a new transaction with the same nonce but higher gas (in MetaMask, the 'speed up' button does this). The distinction matters when reading addresses on Etherscan: the 'Pending Txns' tab shows what's in the mempool; the regular transaction list shows confirmed history.

Builder-extractable value (BEV). As Ethereum's MEV landscape has matured, the value-capture point has moved from searchers (independent bots) to block builders themselves. Builders like Beaver Build, Titan Builder, and rsync-builder run sophisticated MEV-extraction in-house and either capture it for themselves or compete on which builder pays validators the most. This is what proposer-builder separation (PBS) and MEV-Boost were designed to manage. For users, the practical consequence is that even private-mempool submission can leak information to builders who then extract MEV on the order flow.

Solana and other chains. Solana doesn't have a public mempool in the same way — transactions are sent to the current leader (block producer) and either included or dropped. This eliminates traditional front-running but introduces different MEV dynamics: priority-fee auctions, leader spam, and validator-level extraction. Each chain's transaction-submission model has its own MEV profile, and the analysis is chain-specific.

Example

Watch a real DEX swap in two paths. Path A: user broadcasts a $50,000 USDC → ETH swap on Uniswap V3 via MetaMask's default RPC. The transaction lands in the public mempool. A sandwich bot detects it within 50 ms, front-runs with a buy that pushes ETH up 0.4 percent, the user's swap executes at the moved price (paying ~0.4 percent extra), and the bot back-runs with a sell capturing the difference. Net user loss to the sandwich: ~$200. Path B: same user submits the same swap via Flashbots Protect RPC. The transaction goes directly to a builder, never appears in the public mempool, and executes at the unmoved price. Same gas cost, same execution, no sandwich loss. The Flashbots Protect path costs nothing extra and is a one-line MetaMask configuration change. For any meaningful trade size, it's the default-correct option.

Common mistakes

  • Assuming the public mempool is some kind of authoritative state. It's a transient queue; transactions can be dropped, replaced, or reordered.
  • Ignoring MEV for 'small' trades. The threshold at which sandwich attacks become profitable is lower than most users assume — a few thousand dollars on a thin pool is enough.
  • Treating Flashbots Protect as 'paranoid.' It's standard infrastructure used by serious traders for years; the marginal cost is zero and the benefit is real.
  • Forgetting that builders see your transaction even when the mempool doesn't. Private submission reduces but does not eliminate information leakage.
  • Generalising EVM mempool dynamics to Solana or Bitcoin. The submission models are different; the MEV profiles are different.

Check your understanding

You are about to execute a $30,000 DEX swap on Uniswap V3. Which submission method most directly reduces your exposure to sandwich attacks?

Key terms covered

Sources & further reading

We prioritise primary sources. Where a topic moves quickly (regulation, security incidents), we re-check sources on the cadence shown by the page's "Next review" date.

Go deeper