Glossary
Data Availability Sampling
A technique that lets light nodes confirm an entire block's data was published by randomly downloading only a few small pieces, giving strong probabilistic guarantees without full downloads.
Data availability sampling (DAS) lets a node verify that all the data behind a block has been published without downloading the whole thing. Each node requests a handful of small, randomly chosen chunks; if every request succeeds, it is statistically near-certain that the full data is recoverable.
The trick relies on erasure coding, which expands the data so that any sufficiently large subset can reconstruct the original. A block producer that withholds data must hide a large fraction of the chunks, which random sampling will almost always catch. Paired with KZG commitments to bind each chunk to the block, DAS lets resource-light clients help secure data availability and is the cornerstone of danksharding's scaling plan.