Provably-fair · pay-per-use NockApp · demo

Forfeit Dice

Provably-fair, provably-uniform dice — even money over a fixed line.

Trust badge: verified

Outcome construction (public)

roll = rejection-sampled H(serverSeed ‖ clientSeed ‖ nonce) mod 10000 → player wins iff roll ≥ 5000

Forfeit Dice — round #0

demo · no real NOCK

1House commits

The house locks a secret serverSeed, publishing only its hash. The peek surface is commitment-only — never the seed.

2Player commits

The player locks their own clientSeed before any reveal, so neither side can steer the roll.

Reveal & recompute

Complete both commitments, then reveal — the roll is a rejection-sampled H(serverSeed ‖ clientSeed ‖ nonce) mod 10000, recomputed live in your browser.

Provable uniform distribution

Inject bias (negative control)

Recompute 5,000 deterministic rolls in your browser and run a chi-square goodness-of-fit (10 buckets, df=9). Toggle the negative control to confirm the test actually catches a biased stream.

Earned trust badge

Signed Ed25519 issuance receipt over the lab report snapshot 28d565bae5f10035. Freshness: fresh.

Verify badge

Launch evidence

Readiness verified · score 96. Lab report → case → readiness report → badge.

Open case

Settlement residual (honest)

Fairness and forensics are fully trustless. Settlement is bonded-cooperative + timeout-forfeit (HTLC-grade), not contract-grade forced payout — Nockchain consensus cannot gate a payout on the game outcome. Disclosed in docs/FORFEIT-DICE-DOSSIER.md.

This is the fairness and forensics showcase: the commit-reveal flow and the verifier run entirely in your browser and no real NOCK changes hands. Real-NOCK play is gated behind a house wallet, fakenet end-to-end testing, and the disclosed settlement model.