Sequencer

On most layer 2 networks, users send transactions to a sequencer rather than to a peer-to-peer mempool. The sequencer orders them, executes them into L2 blocks within milliseconds or seconds, and gives users a quick soft confirmation. Periodically it compresses the blocks and publishes them on Ethereum, usually as blobs, where they become part of the rollup's verifiable history.

The sequencer is a major source of revenue and power. It collects L2 fees, pays for L1 data and keeps the difference, and it decides ordering, which means it can capture or auction MEV. Most large rollups still run a single sequencer operated by the founding team, which makes it a key point in any decentralisation assessment.

A centralised sequencer can go down, halting block production, or delay specific transactions. It cannot steal funds or forge invalid state if the rollup has working proofs, and well-designed rollups let users force transactions through L1 if the sequencer censors them. Shared and based sequencing are proposals to reduce this dependence.