Taiko, a based rollup, recently completed post-incident upgrades and bridge recovery, putting a spotlight on the real-world tradeoffs of this sequencing design. Based rollups ditch the dedicated L2 sequencer and instead let Ethereum's L1 block proposers order transactions — a move that reduces one central point of control but introduces a fresh set of complexities around MEV, latency, and relay concentration.
How based sequencing works
In a based rollup, a user submits a transaction to the L2 mempool or a gateway. A builder bundles those L2 transactions into an EIP-4844 blob, and the L1 proposer picks the winning block under PBS. L2 nodes then watch the L1 chain and derive the next L2 block from that order. Proofs are posted separately. The whole process runs on Ethereum's 12-second slot rhythm, which sets the heartbeat for L2 ordering finalization.
That rhythm is a double-edged sword. The L2 inherits Ethereum's liveness and proposer set — no single sequencer can halt the chain or censor transactions. But the UX is tied to L1 conditions: users face the same 12-second cadence, rely on preconfirmations for speed, and occasionally deal with reorg risk.
The centralization that remains
Based rollups reduce one form of centralization, but they don't eliminate it. Builders and proposers still decide what goes in, and PBS relays can become informal gatekeepers. If a handful of relays or builders dominate, they become a soft chokepoint — even without a centralized sequencer inside the L2.
MEV also migrates. Under PBS, L1 builders and proposers capture the value, and L2 preconfirmations become a market. Cross-domain MEV lets searchers compose L1 and L2 opportunities inside the same L1 block, concentrating edge in the best-equipped builders. That raises questions about revenue splits: who captures L2 MEV? Some designs sketch kickbacks to the L2 or app-level rebates, but nothing is settled.
What Taiko's upgrade shows
Taiko's post-incident upgrades and bridge recovery highlight how governance and proof systems interact with based designs. The incident itself isn't detailed, but the response underscores a key point: based rollups are not a replacement for proofs. Sequencing decentralization does not verify state transitions. Security still depends on the proof system, data availability, and bridges — and based sequencing does not fix weak proofs or leaky bridges.
The design also doesn't guarantee perfect fairness. Builders and proposers still pick what goes in, and PBS/relays influence the flow. Based rollups are not automatically faster either; they ride Ethereum's slot cadence and congestion.
Unresolved questions
The biggest open question is who gets the MEV. Without a clear revenue model, L2s might struggle to fund themselves. And while based rollups offer a cleaner neutrality story than a centralized sequencer, they shift the concentration risk to the relay and builder layer — a problem that PBS governance hasn't fully solved.
Taiko's next steps will be watched closely. The upgrade shows that even with a based design, incidents happen and governance matters. The real test is whether the tradeoffs — UX latency, MEV migration, relay gravity — are worth the gain in liveness alignment.
