Data Availability Sampling (DAS)

Overview

How it works

Data Availability Sampling reduces the burden of checking large block or blob data. Instead of downloading everything, many nodes request randomly selected pieces and verify them against a cryptographic commitment. Erasure coding adds redundancy so withholding any unrecoverable portion requires a large fraction of encoded pieces to be unavailable. Repeated random sampling then makes concealment increasingly likely to be detected, providing a probabilistic availability guarantee while preserving lightweight participation.

Implementations must verify commitments, sampling randomness, peer responses, coding parameters, and enough independent samples. DAS establishes that data was available, not that transactions were valid or executed correctly. Rollups and light clients still need the appropriate validity, fraud, or consensus verification for state transitions.

What matters in practice

DAS lets many light participants sample small, randomly selected portions of erasure-coded data instead of downloading the entire dataset. When designed correctly, wide independent sampling makes wholesale withholding harder to conceal while keeping verification affordable. The scheme still needs sound coding, commitments, network distribution and assumptions about the number of independent samplers. Passing sampled checks is probabilistic confidence in availability, not proof that a smart contract executed correctly. For payments, state reconstruction and exit options matter as much as throughput claims.

The practical security question is how much independent sampling actually occurred. A single operator requesting a few fragments does not provide the same assurance as thousands of independently chosen samples with sound encoding proofs. Check whether the data commitments are cryptographically binding and whether a user can still obtain enough data to reconstruct the relevant block after the nominal availability window.

Related glossary concepts

Data Availability