From "Block" to "Blob," the evolution speaks volumes.
Executable "shard chains" with "crosslinks" were phased out: EVM implementation in the Beacon Chain, a rollup-centric Ethereum roadmap leveraging "data availability sampling" to scale the base layer without increasing application complexity. But how do you invoke sharded content without block metadata?
Enter "blobs." The term "Blobspace" couldn’t be more fitting!
The Evolution of Ethereum Sharding
Sharding (or "Phase 1") was originally slated to follow "Phase 2" (Beacon Chain execution environments). However, priority shifted to mainnet EVM before "Phase 0" (Beacon Chain launch), while "Phase 2" execution layers (e.g., eWASM) remained distant.
The "Phase 1" spec underwent multiple rewrites pre-Beacon Chain:
- Fewer shards (1024 → 64)
- Idealized cross-shard communication (crosslinks) enabling free-riding
- New custody-proof designs (replacing custodial logic with intentional rare attestation losses)
Earlier sharding research was even more abstract and ambitious: cross-domain messaging, eWASM execution environments, statelessness with dynamic access, shard committees—all adding L1 complexity just as L1 began ossifying.
The Pivot to Data-Centric Scaling
But if L1 focused solely on data availability, most complexities shifted to L2 development. Sampling solved L1’s data problem. What if the network layer could support additional features?
On October 14, 2020, developers held a call on "Phase 1 networking." Key takeaways: Gossipsub was hot, while DHTs seemed slow. Yet, this was premature—every network dev was scrambling for the Beacon Chain launch (December 1!).
Biases at the time:
- Gossipsub: Battle-tested, nearly mainnet-ready (aside from DoS fixes).
- Discv5: Incomplete, requiring live migration from v5.0 to v5.1.
👉 Explore Ethereum's P2P Evolution
The Rollup-Centric Roadmap
Direction crystallized: Reduce L1 complexity. The Beacon Chain was enough. Scale via data—long-term through data availability sampling (DAS)—and embrace L2 solutions. Vitalik framed this as A Rollup-Centric Ethereum Roadmap.
Post-launch, researchers drafted early DAS designs while implementers tackled testnets (Zinken, Pyrmont) and bugs. Network-layer designs emerged, though DAS overshadowed all.
Enter "Blobs"
The term "blob" debuted in sharding specs circa 2021, describing "shard data blobs." Early drafts (like this one) formalized blob typing and networking layers.
By mid-2021, complexity gave way to modularity: Proposer-centric designs simplified sharding into a network-layer challenge. Rollups (especially Optimism) stole the spotlight.
Proto-Danksharding: EIP-4844
In 2022, calls to reduce calldata costs (EIP-4488/4490) clashed with sustainability concerns. Meanwhile, Vitalik and Dankrad refined sharding designs: Beacon Chain-centric, data-only, sampling-focused.
"Proto-Danksharding" (EIP-4844) was born at EthDenver 2022:
- @lightclients and I worked on Geth.
- @asn_d6 advanced KZG research.
- @adietrichs modeled fee markets.
- Prysm built the first CL prototype.
Now, 4844 lays groundwork for full danksharding, but "Blobspace" is the true meme—closer to Ethereum’s vision than prior iterations.
👉 The Future of Ethereum Scaling
FAQs
Why shift from execution shards to data shards?
Execution shards added L1 complexity without commensurate benefits. Data shards empower L2s while keeping L1 lean.
How does DAS improve scalability?
Sampling allows nodes to verify large datasets without downloading them entirely, enabling exponential data capacity growth.
When will danksharding launch?
EIP-4844 ("Proto-Danksharding") is the first step. Full danksharding awaits further research and network upgrades.
Key Takeaways
- Simplicity wins: L1 scales via data; execution migrates to L2.
- Blobs are foundational: EIP-4844 unlocks rollup-friendly bandwidth.
- Future-proof: DAS and danksharding ensure long-term scalability.
Ethereum’s sharding journey—from abstract ambitions to pragmatic "Blobspace"—reflects its ethos: Build, iterate, simplify.