Committee Election

Origin: composes, a VRF-based selection technique popularized by Algorand, not a Solidus original. Solidus status: testnet: the selection mechanism runs live; the target committee scale (21 of up to 100) has not been demonstrated.

Committee election is how a blockchain decides, round by round, which validators get to propose and vote on the next block, rather than fixing the order in advance (predictable, and gameable by an attacker who knows who's next) or picking round-robin (simple, but the same problem). A common alternative is to use a Verifiable Random Function (VRF): each eligible validator computes a value from their private key and a shared, unpredictable seed (usually the previous block's hash), and the outcome determines whether, and in what order, they're selected. Crucially, anyone can verify the selection was fair after the fact, without being able to predict or influence it beforehand.

Where it comes from

This specific technique, VRF-based, verifiable random selection of a rotating committee, was popularized by Algorand's 2017 "cryptographic sortition" design (Gilad, Hemo, Micali, Vlachos, and Zeldovich), and similar ideas appear in Cardano's Ouroboros Praos and other proof-of-stake protocols. Solidus composes the general idea rather than inventing a new selection technique; its own VRF is built on Ed25519, implemented from scratch rather than reused from another chain's codebase.

What's real today versus what's specified

The mechanism itself is genuinely running: on the public testnet, an Ed25519-based VRF, seeded from the previous round's output, picks the block proposer every round. A simpler round-robin mode exists in the codebase purely for local development and is switched off in the deployed network, so this isn't a case of a fallback masquerading as the real thing in production.

What hasn't happened yet is scale. The target design elects a 21-validator active committee out of an eligible pool of up to 100 validators per subnet: a meaningful subset selection, the kind that matters for both security (an attacker can't easily predict who'll be on the committee) and performance (not every eligible validator has to process every round). The live testnet has exactly four validators total, so there's no larger pool to select a subset from: the VRF still runs and still determines the proposer, but "election" at n=4 doesn't yet demonstrate the committee-scaling behavior the design is built for.

Check it yourself

The VRF leader-selection function, and the flag that disables it for local-only development, are both visible in source in the public solidusnetwork/protocol repository's consensus crate, no running node required to confirm the mechanism exists as described.

Related terms: HotStuff · BFT · Validator · Proof-of-Identity · Finality · VRF · Ed25519