The rollup era begins

By 2021, Ethereum's success was its problem: demand outstripped block space and gas spiked. Rather than scale the base layer by making it heavier, Ethereum committed to a rollup-centric roadmap — push execution to L2s, keep L1 as the secure settlement and data layer.

What a rollup is

A rollup runs transactions off-chain, then posts the results (and the data to reconstruct them) back to Ethereum, inheriting L1 security. Two flavors emerged: optimistic rollups (assume valid, allow fraud proofs during a challenge window — Arbitrum, Optimism) and ZK rollups (prove validity with a cryptographic proof).

        users ──→  L2 (Arbitrum / Optimism)   ← cheap, fast execution
                       │ posts data + results
                       ↓
                   Ethereum L1                 ← security + data availability

Optimistic vs ZK, in one breath

Optimistic rollups assume every batch is valid and let anyone post a fraud proof during a challenge window (about a week) to claw back a bad state — cheap to run, but L1 withdrawals wait out the window. ZK rollups attach a validity proof to every batch, so correctness is proven up front and withdrawals are fast — but the proving is computationally heavy. The industry has been marching from the first toward the second ever since.

OPTIMISTIC   batch ─→ [ ~7-day challenge window ] ─→ final
              "valid unless someone proves fraud"

ZK           batch + validity proof ─→ final (fast)
              "proven correct on arrival"

Both flavors share one bottleneck, and it's the one the whole roadmap kept circling back to: they must publish their data to L1 so anyone can reconstruct and verify, and L1 data was expensive. That single constraint is why everything after — 1559's fee market, the Merge's security, and ultimately 4844 blobs — mattered so much to L2 economics.

It worked. The roadmap that started here is the through-line for everything since: 1559 priced L1 better, the Merge secured it with stake, and 4844 blobs made the L2 data cheap. "Scaling" stopped being a promise and became an address you bridge to.