EIP-4844: Shard Blob Transactions, colloquially known as Proto-Danksharding, is an Ethereum Improvement Proposal authored by Vitalik Buterin, Ansgar Diederich, Cody Inabinet, Hsiao-Wei Dietrichs, and Mofi Storr. It was activated in the Dencun hard fork on March 13, 2024. The name pays tribute to researcher Dankrad Feist, whose full danksharding proposal is the long-term data availability road map that EIP-4844 partially implements.

EIP-4844 introduced a new type of Ethereum transaction that carries blobs — large chunks of raw binary data (~128 KB each). Blobs are priced separately from calldata, stored by consensus nodes for only ~18 days (not indefinitely), and committed to via a KZG polynomial commitment rather than a hash. The immediate practical effect was to reduce rollup costs by 10–100x.

> Source: The EIP is at eips.ethereum.org/EIPS/eip-4844. The companion danksharding design is at notes.ethereum.org/@dankrad/new_sharding.

---

Publication and Context

Before Dencun, Ethereum rollups (Optimism, Arbitrum, zkSync, Starknet, Base) posted their transaction data as calldata in Ethereum transactions. Calldata is permanent (stored forever by all full nodes) and priced for storage permanence. This made it expensive — the dominant cost for rollups posting to Ethereum was this calldata fee, which meant L2 users paid $0.10–$5 per transaction even at moderate activity.

The core insight: rollups only need their data to be available for ~1–2 weeks — long enough for fraud proofs (Optimistic rollups) or proof submission (ZK rollups). There is no reason to pay for permanent storage.

EIP-4844 created a new lane for this temporary data: blobs.

---

Blobs: The Core Mechanism

A blob (binary large object) is a ~128 KB chunk of data attached to a special blob-carrying transaction:

Property	Value
Size per blob	~128 KB (4096 field elements of 32 bytes each)
Max blobs per block	6 (target: 3)
Data retention	~18 days (4096 epochs) by consensus nodes
EVM access	EVM cannot read blob contents directly — only their KZG commitments
Fee market	Separate blob fee market with its own base fee and EIP-1559 adjustment

Blobs are committed to via KZG polynomial commitments (Kate-Zaverucha-Goldberg commitments). This is a cryptographic commitment scheme with a useful property: given a commitment, you can prove that a specific piece of data is indeed the evaluation of the committed polynomial at a specific point. This is the building block for Data Availability Sampling in full danksharding.

---

Why KZG Commitments?

A simple Merkle hash of the blob data would work for commitment, but KZG provides additional advantages:

1. Efficient proofs: KZG allows constant-size proofs that a specific element is present in the committed data — no Merkle path overhead
2. Polynomial structure: The blob data is treated as coefficients of a polynomial over a finite field; the commitment is an elliptic curve group element
3. Danksharding readiness: In full danksharding, KZG commitments enable Data Availability Sampling — light nodes can request random polynomial evaluations and verify them, checking availability without downloading the full blob

The KZG ceremony (also called the “Powers of Tau”) was a trusted setup ceremony completed in 2023 with over 140,000 participants — generating the structured reference string (SRS) needed for KZG proofs.

---

Economic Impact

Before Dencun (calldata-based):

• Posting 100KB of data: ~100,000 gas × 30 gwei ≈ $8–15 on a busy day
• Optimism and Arbitrum: $0.10–$3.00 per L2 transaction

After Dencun (blob-based):

• Blobs have a separate, independent fee market
• Base blob fee starts very low; rises only if blocks are consistently full of blobs
• Early Dencun: L2 fees dropped to $0.001–$0.05 per transaction

The blob fee market follows the same EIP-1559 adjustment mechanism as regular gas but is completely independent — L1 calldata congestion doesn’t directly raise blob fees.

---

The Path to Full Danksharding

EIP-4844 is “proto” because it does not implement full danksharding. The differences:

Feature	Proto-Danksharding (EIP-4844)	Full Danksharding
Blob count	3 target, 6 max per block	64+ blobs per block
Data availability	Nodes download all blobs	DAS: nodes sample random blobs
Throughput	~0.375 MB/block target	~16+ MB/block target
Light client support	No DAS	Full DAS with KZG proofs

Full danksharding requires a distributed validator network (DVT), Data Availability Sampling by light clients, and further engineering — expected in a future Ethereum upgrade.

---

Reality Check

Dencun launched smoothly and delivered on its fee reduction promise. The blob market has occasionally filled (blob fees spiking during high activity), but average fees remained dramatically lower than the calldata era. The main ongoing limitation is the 6-blob-per-block cap — during periods of high rollup activity, competition for blob space emerges. The long-term path to full danksharding, with proper DAS, remains years away.

---

Related Terms

• EIP-4844
• Data Availability
• KZG Commitment
• EIP-1559 Whitepaper
• Celestia Whitepaper

---

Research

• Buterin, V., et al. (2022). EIP-4844: Shard Blob Transactions. eips.ethereum.org.

— The EIP itself. The specification (§Specification) and design rationale are the primary reference.

• Feist, D., & Khovratovich, D. (2020). Fast Amortized Kate Proofs. Research proposal.

— Introduces efficient KZG proof aggregation that makes blob proving practical at scale.

• Boneh, D., Drake, J., Fisch, B., & Gabizon, A. (2020). Efficient Polynomial Commitment Schemes for Multiple Points and Polynomials. IACR ePrint 2020/081.

— KZG commitment scheme mathematics that underlies EIP-4844’s proof structure.