What is a based rollup?

Defining what is and what isn't a based rollup.

There is no consensus definition of a based rollup. Rough consensus has shifted since publishing, but Justin's original post uses this definition:

A rollup is said to be based, or L1-sequenced, when its sequencing is driven by the base L1. More concretely, a based rollup is one where the next L1 proposer may, in collaboration with L1 searchers and builders, permissionlessly include the next rollup block as part of the next L1 block.

The key to this definition is that sequencing is "driven" by a base layer and not controlled by a completely external party. The rest of the definition after "more concretely" has materially shifted with more teams modifying their sequencing rules for based preconfs. In fact, the definition has expanding in complexity. In the rest of this document, we will attempt to set up a reasonable definition that maintains the spirit of Justin's original post while being flexible enough to be practically useful.

Justin's definition (especially "rollup block as part of the next L1 block") implies that based rollup block proposals are embedded within an L1 block as L1 transactions. It follows, then, that a based rollup's canonical ordering can be derived by looking at the contents and ordering of L1 blocks. From this, we can deduce that a based rollup's finality is inherited from the base layer (if we assume all based rollup blocks are valid if proposed). We must also notice that reorgs or downtime in the base layer's consensus will also affect the based rollup.

The "driven" part of the definition is a bit harder to expand on. We find it useful to look at the goals of a based rollup, one of which is enabling synchronous composability between the base layer and the based rollup during block proposal. For this to be possible, some communication or coordination must take place between the block production processes of the base layer and the based rollup. Justin envisioned this coordination as taking place internally within one block builder entity, but the goal of synchronous composability doesn't require that coordination means strictly internal communication. Other forms of communication (such as trusted orderflow sharing using a network of TEE builders) could be adapted to this cause.

Combined, these two properties set us up for a rather straightforward definition of based sequencing:

---

Based sequencing is a system for sequencing (ordering transactions and getting them into blocks) that must satisfy the following criteria:

1. coordination: The base layer (explicitly it's blockspace, but in practice the entities with sequencing rights to that blockspace - L1 proposers in the lookahead on Ethereum) are granted an advantage in the sequencer election that determines which "sequencer" has a write lock to the rollup. This advantage can be complete (called total anarchy based sequencing) or incomplete (i.e. running an onchain auction to distribute sequencing rights that gives an explicit last look to base layer proposers).

2. shared settlement: The based rollup reorgs with and reaches finality with the base layer. This means that if the base layer reorgs, a based rollup will reorg. This is implemented through atomic block proposals, where a block proposal for the based rollup is represented as a base layer transaction and the canonical sequence of the based rollup is derived from these base layer transactions.

A based rollup is a rollup (see also here) that uses primarily based sequencing. It may use a fallback.

A based L2 is an L2 that uses primarily based sequencing. Note that a based L2 might not use Ethereum for DA and thus would not be considered a rollup.

---

See also: