For most of the past year, layer-2 networks competed on execution environments and fee markets while quietly diverging on the one component users notice most when it fails: transaction ordering.
That divergence appears to be narrowing. Specifications published over the past fortnight by three of the largest rollup teams describe broadly compatible approaches to shared sequencing, in which ordering is delegated to a network of participants rather than a single operator.
The practical benefit is atomic composability across chains — the ability to submit a transaction that either succeeds on two networks or fails on both. Today, bridging between rollups requires trust assumptions that vary by route.
Engineers involved in the drafts describe the convergence as pragmatic rather than coordinated. Each team faced the same constraint: censorship resistance is difficult to argue for when a single sequencer can reorder or delay transactions.
Open questions remain around economics. Shared sequencing redistributes fee revenue that currently accrues to individual operators, and no team has published a settled model for how that revenue would be split.
Testnets implementing the drafts are expected in the fourth quarter. Production deployments, participants said, are unlikely before next year.
