Data Unavailability Penalty

A Data Unavailability Penalty is a cryptoeconomic slashing mechanism that punishes sequencers or validators who withhold the transaction data necessary for fraud or validity proofs. This ensures that the state of a rollup or modular chain can be independently verified by anyone, maintaining security and trustlessness.

• Core Purpose: Prevents validators from publishing only state commitments while hiding the data, which would make fraud undetectable.
• Trigger Condition: Penalties are applied when a node challenges missing data and the data is not provided within a challenge period.

The penalty is intrinsically linked to a dispute or challenge period. After a block is proposed, a verifier has a fixed window (e.g., 7 days) to request the underlying transaction data.

• If the data is provided, the chain proceeds normally.
• If the data is withheld and not provided by the end of this window, the assumption of malfeasance is made, and the slashing penalty is automatically executed.
• This creates a strong incentive for data publishers to keep data available and retrievable.

The penalty typically involves the confiscation (slashing) of a portion of the validator's staked assets or bond. This serves two key functions:

• Punishment: Makes withholding data economically irrational, as the slashed amount far exceeds any potential gain from an attack.
• Compensation: The slashed funds can be burned or redistributed to honest participants, aligning economic incentives with network security.
• The severity is often a significant percentage of the total stake, making attacks prohibitively expensive.

This penalty is fundamental to modular blockchains and optimistic rollups, which separate execution from consensus and data availability.

• In Celestia and EigenDA, it secures the data availability layer itself.
• In Optimism and Arbitrum, it ensures the sequencer posts data to L1 (Ethereum) so fraud proofs can be executed.
• Without this penalty, a rollup could become a centralized, non-verifiable system if its operator stops cooperating.

It's crucial to distinguish the penalty from the challenge mechanism that triggers it.

• Data Availability Challenge: A technical process where a light node samples data and issues a challenge if blocks are missing. This is the detection mechanism.
• Data Unavailability Penalty: The economic consequence (slashing) applied after a successful challenge. This is the enforcement mechanism.
• Together, they form a complete detect-and-punish security model.

EIP-4844 introduces blob-carrying transactions to scale data availability for L2s. While Ethereum itself does not directly slash for data withholding, it creates the economic framework.

• Rollup sequencers pay a fee to post data blobs to Ethereum.
• Blobs are deleted after ~18 days, but must be available for the fraud proof window.
• The cost of withholding data is the loss of all fees and the security failure of the L2, which is enforced by the L2's own penalty mechanism and market forces.

A Data Unavailability Penalty is a cryptoeconomic slashing condition enforced on a blockchain's consensus layer. It is triggered when a block producer (e.g., a rollup sequencer or Layer 1 validator) successfully proposes a block but withholds the corresponding transaction data. The penalty typically involves the confiscation (slashing) of a portion of the offender's staked assets and may include ejection from the validator set. This creates a strong financial disincentive against data withholding attacks.

In optimistic and zk-rollup architectures, this penalty is the primary enforcement for the data availability requirement. Sequencers must post transaction data to the base layer (Layer 1). If data is unavailable, verifiers cannot reconstruct the rollup state or compute validity proofs, breaking the security model. Penalties ensure sequencers are financially accountable, making the system trust-minimized. Without this, users could be forced to trust the sequencer's honesty.

The penalty is often linked to a challenge-response protocol. After a block is proposed, there is a dispute window (e.g., 7 days in optimistic rollups) during which any watcher can submit a fraud proof if data is missing. The penalty is executed upon successful verification of the challenge. This design leverages economic games and decentralized watchtowers to keep the system secure without requiring constant active participation from all users.

This penalty is distinct from other consensus slashing conditions:

• Double-Signing Penalty: Punishes proposing conflicting blocks.
• Liveness Penalty: Punishes being offline.
• Data Unavailability Penalty: Specifically punishes the withholding of published data, not its correctness. It addresses a unique data withholding attack vector where a malicious actor publishes only a block header, making state transitions unverifiable.

• Ethereum's EIP-4844 (Proto-Danksharding): Introduces data availability sampling and slashing for data availability committees in future sharding designs.
• Celestia: A modular blockchain network specializing in data availability, with penalties for data withholding by rollup sequencers using its base layer.
• Polygon Avail: Implements a KZG commitment scheme and validator slashing to guarantee data is published and available.

• Data Availability Sampling (DAS): A technique where light nodes randomly sample small pieces of block data to probabilistically verify its availability, enabling scalable verification.
• Data Availability Committee (DAC): A trusted group that signs attestations confirming data is available; a weaker, more centralized alternative to on-chain penalties.
• Validity Proof: In zk-Rollups, a cryptographic proof that state transitions are correct, which requires the underlying data to be available for proof generation and verification.

EigenDA, a restaked data availability layer on EigenLayer, employs a dual-quorum system. Operators commit to storing and serving data blobs. Dispersal nodes send data to operators, and Availability nodes verify its correctness. Operators who fail to store data, serve it upon request, or attest incorrectly are subject to slashing of their restaked assets, directly penalizing unavailability.

In Arbitrum's AnyTrust chains (like Nova), a Data Availability Committee (DAC) is trusted to make data available. While the core protocol lacks a native crypto-economic penalty, it incorporates a challenge period. If the DAC fails, users can challenge the state, forcing a fallback to slower, fully on-chain data availability. This creates a strong disincentive for committee misbehavior to avoid protocol failure.

Similar to Celestia, Polygon Avail uses erasure coding and data availability sampling. A network of fishermen (full nodes) monitor the chain. If a block producer withholds data, fishermen can submit a fraud proof demonstrating the data is unreconstructible. Upon verification, the malicious validator's stake is slashed, providing a clear economic penalty for data unavailability.

A compact cryptographic proof that demonstrates a state transition executed in a rollup or optimistic rollup is invalid. Fraud proofs are only possible if the underlying transaction data is available for anyone to reconstruct and challenge the state.

• Dependency on Data: A data unavailability attack directly neutralizes fraud proofs, as verifiers cannot compute the correct state to find the fraud.
• Contrast with Validity Proofs: Validity proofs (ZK-proofs) do not have this dependency, as the proof itself guarantees correctness.

A trusted, permissioned set of entities that sign attestations confirming they have received and stored the data for a block. This provides a weaker trust assumption than full on-chain availability but is simpler to implement.

• Role: Acts as a fallback data source if the primary layer fails.
• Limitation: Relies on the honesty of committee members, introducing a trust vector that pure cryptographic solutions like DAS aim to eliminate.

A specialized blockchain layer whose primary function is to order and guarantee the publication of transaction data for other execution layers (like rollups). It decouples data availability consensus from execution.

• Examples: Celestia, EigenDA, and Avail.
• Purpose: Provides a scalable, cost-effective base for rollups to post their data, with its own security and consensus focused solely on data availability.

The malicious act by a block producer (sequencer, validator) of publishing a block header but not making the corresponding transaction data available. This is the specific failure mode that a data unavailability penalty is designed to slash.

• Goal: To force the network to accept an invalid state transition or to censor transactions.
• Mechanism: The attacker exploits the delay between block proposal and the requirement for data to be available for verification.