BLS Signature

A BLS signature is a digital signature with one unusual, very useful property: many separate signatures on the same message, from many different signers, can be squashed ("aggregated") into a single signature that's exactly as small as one individual signature, and still verifies against all the signers' public keys at once.

That property matters enormously for blockchain consensus. A committee of validators voting to finalize a block each produces its own signature; without aggregation, a block certificate has to carry every individual signature (hundreds of them, for a large committee) or expensive proofs of who voted. With BLS, the block carries one short aggregate signature plus a bitmap of who's included, regardless of whether five validators voted or five hundred. This is why BLS is the default choice in Ethereum's beacon chain, Chia, Algorand, and most modern BFT ("Byzantine Fault Tolerant") consensus designs, including Solidus's own.

The aggregation works because of the underlying math: BLS signatures live on a pairing-friendly elliptic curve (Solidus uses BLS12-381, the same curve family BBS+ credentials build on), where a "bilinear pairing" operation lets you check a relationship between multiple signatures and multiple public keys in one step, instead of verifying each pair individually.

Where it comes from

Designed by Dan Boneh, Ben Lynn, and Hovav Shacham in their 2001 paper "Short Signatures from the Weil Pairing." Twenty-plus years of subsequent cryptanalysis and adoption across major blockchain networks is what makes it a safe default choice today. Solidus did not design any part of it.

Solidus status

Solidus's HotStuff-based consensus layer uses BLS signature aggregation so that validator committee votes collapse into a single compact certificate per block, rather than one signature per validator. This shipped as part of the protocol's multi-node consensus milestone, alongside VRF-based committee election (see that entry) and libp2p networking, with test coverage in the protocol's Rust test suite.

Two things must not get blurred here. First, the design targets a 21-of-100 validator committee, elected via VRF; the running testnet is a 4-node dev committee. That's a real gap between spec and reality, and it should be stated as such every time, not smoothed over. Second, no independent audit of this consensus code, BLS aggregation included, has happened, and there is no mainnet.

Proof: honestly, there isn't a stranger-runnable demo of BLS aggregation specifically today, it's internal machinery inside vote and certificate formation, not something exposed as a queryable endpoint. The adjacent, independently-checkable evidence is the consensus algorithm's correctness work more broadly: an exhaustive TLA+ model check (963,033 states explored, zero safety violations found) and a differential-oracle reproduction an anonymous stranger ran from a fresh clone. Neither is BLS-specific proof, and it would be dishonest to present them as one.