Rollups and fragmentation

Wonder

1. How can a chain process thousands of transactions per second while relying on a base layer that processes only a fraction?
2. What specific trust assumptions replace cryptographic proofs in optimistic rollups, and who bears the risk if those assumptions fail?
3. Why does moving liquidity to lower fees simultaneously fragment it across dozens of isolated chains?

Explore

1. Ethereum’s scaling strategy shifted execution off mainnet. Rollups like Arbitrum, Optimism, and Base process transactions independently but post transaction data and state roots back to Ethereum. This preserves the security properties of the base layer while allowing higher throughput. The fixed cost of posting data is amortized across batches of user transactions, drastically reducing per-transaction fees compared to executing directly on mainnet.
2. Optimistic rollups assume validity by default. They post transaction data without proofs, relying on a challenge period, commonly seven days, during which anyone can submit a fraud proof if an invalid state root is proposed. Zero-knowledge rollups differ fundamentally: they generate cryptographic proofs of computation correctness before posting. This allows for faster finality but requires complex proving hardware and recursive proof composition to keep the overhead manageable.
3. Fragmentation is the primary economic cost of this architecture. Liquidity splits across dozens of chains, requiring users to bridge assets between them. Bridges act as intermediaries that lock tokens on one chain and mint equivalents on another. This complexity introduces new attack surfaces; a compromised bridge contract can drain funds from multiple ecosystems simultaneously, creating systemic risk that did not exist in the monolithic era.
4. Centralization persists at critical points despite decentralization claims. Sequencers order transactions off-chain before batches reach mainnet. Currently, most rollups operate with centralized sequencers controlled by their respective teams. These entities hold upgrade keys and can pause or modify protocol parameters. L2BEAT rates rollups by stage, from Stage 0, where operators can change the system at will, to Stage 2, where users can always exit and upgrades are delayed or limited.
5. The Dencun upgrade (March 2024) introduced EIP-4844, providing cheap ‘blob’ space for rollup data. This reduced the cost of posting transaction calldata to mainnet, making rollups more economically viable. However, it also increased competition among chains for user attention and liquidity. The result is a crowded ecosystem where fee advantages are temporary and user migration is frequent, driven by incentives rather than deep integration.
6. Open L2BEAT and examine the ‘Stages’ tab. Compare a Stage 0 chain with a Stage 2 chain. Note which operators hold sequencer keys and what governance mechanisms exist for upgrades. Then visit Etherscan and inspect a recent rollup batch transaction; observe the data size posted to mainnet versus the number of user transactions included in that batch. This reveals the compression ratio driving fee reductions.

Key facts

Exercise nine, six, three, one

Fact check

Dates, figures and claims checked against the published sources listed below.

• L2BEAT, 'Stages Framework' (accessed 2024)
• Ethereum Foundation, 'Dencun Mainnet Announcement' (March 2024)
• ethereum.org, 'Optimistic Rollups' and 'ZK-Rollups' documentation