VRF

A VRF is a function that takes a private key and an input, and produces two things: a pseudorandom output that looks like noise to anyone without the key, and a proof that lets anyone holding the matching public key verify the output was computed correctly, without being able to predict it in advance or fake a different output after the fact.

The use case that makes VRFs load-bearing in modern blockchain design is leader and committee election. A consensus protocol needs to pick, each round, which validator (or which small committee) gets to propose the next block, and it needs that pick to be unpredictable in advance (so an attacker can't target the next leader before they're chosen) but verifiable after the fact (so everyone else can confirm the selection was legitimate, not rigged). A VRF does exactly this: each validator runs the VRF locally with the round number as input; whoever's output falls below some threshold is elected, and they can prove it to everyone else with the VRF proof, without anyone needing to trust a central coordinator or run an expensive multi-party coin flip.

Where it comes from

VRFs were formalized by Silvio Micali, Michael Rabin, and Salil Vadhan in 1999. The IETF's Crypto Forum Research Group later standardized a concrete, interoperable construction (draft-irtf-cfrg-vrf), edited by Sharon Goldberg, Leonid Reyzin, Dimitrios Papadopoulos, and Jan Vcelak. Algorand, Cardano's Ouroboros Praos, and Chia all use VRFs for leader election in essentially this way. Solidus uses the same tool for the same purpose: the VRF primitive is not something Solidus built.

What is new is one layer up, and it's the one thing in this whole Lexicon marked invented rather than composes: Solidus's consensus design gates committee eligibility, who is even allowed to be a candidate the VRF can select from, on a verified human-identity credential, not on staked capital. The VRF still does the same unpredictable-but-provable selection job it does everywhere else; what changes is the eligible set it's selecting from. See Proof-of-Identity for that part of the design.

Solidus status

Implemented and running: Solidus's HotStuff-based consensus uses VRF-based election, shipped as part of the multi-node consensus milestone alongside BLS signature aggregation (see that entry) and libp2p networking, with test coverage in the protocol's Rust test suite.

The honest gap: the design specs a 21-of-100 validator committee. The running public testnet is a 4-node dev committee, at that scale, VRF election isn't really doing the job it was designed for yet (selecting a small committee out of a large, contested pool). That's a real, stated distance between the design and what's running today, and it should never be blurred in either direction. No independent audit has been performed, and there is no mainnet.

Proof: none exists today that isolates VRF-based election specifically for a stranger to run and check, say that plainly rather than reach for an adjacent number and imply it covers this. The consensus algorithm's broader correctness work (TLA+ model checking, a stranger-reproduced differential-oracle test) is real and checkable, but it is not VRF-specific proof.