Decentralizing skill-based fun

Every online prize game today rests on one assumption: that the operator scored you honestly and will pay you. That's a lot of trust to extend to a stranger holding a pot of money. Replace the operator's word with a re-runnable computation and most of that trust requirement evaporates — you don't have to believe the result, you can reproduce it.

Start with the important part: it's a game. Free to play, fun on its own, no wallet required to drop into an arena. The prize layer is opt-in, and it rides on the exact same machinery — because the thing that makes a staked match provable is the same thing that makes a friendly match fair: the engine is deterministic, so the result is a fact, not an opinion. You don't need the money for the math to matter; the money is just one thing the math lets you do safely.

The chain runs zero game logic. A hundred-player match is on the order of ten million ticks; that could never fit in a transaction, and it doesn't try to. Solana plays two narrow roles instead — a bank that holds custody of the stakes, and a notary that timestamps the commitments. The signed re-run off-chain is what proves the game; the chain just holds the money and the receipts.

Stakes sit in an on-chain escrow vault. When the deterministic result is settled, it commits a Merkle root of the payouts, and each winner pulls a recipient-bound claim — a leaf that only their own wallet can redeem. The money follows the proof, not a person: no operator hits "send" on your winnings, you claim them against a root the chain already holds.

Today the operator still computes the payouts off-chain (that's where the ten-million-tick game lives). But it can't cheat quietly, because the chain is armed with guardrails: it caps over-payment with hard invariants, pins an immutable seed and codex root, and opens a challenge window in which anyone can file a cheap mismatch proof — recompute the roots, and if they don't match what was committed, the settlement freezes. Honesty isn't assumed; dishonesty is catchable and expensive.

Here's the part we think is genuinely new. When a game's rules are math and its record is public, the game no longer depends on the people who made it. The engine is open-source C++; the replay format is a documented bit-packed codex; the verifier runs anywhere. Nothing about deciding a winner needs us to be alive, funded, or honest.

So custody can walk down a ladder away from control. Today an operator runs settlement; next it sits behind a multisig; in principle it ends at no one — a keeper set anyone can join, running the same open settler against the same public records. At that point the game is closer to a protocol than a product: it keeps working as long as people want to play it, and no single party can quietly change the rules, take a bigger cut, or switch it off.

This is deliberately unfinished in one respect: real-value play is disabled while the proof matures. The founding edition runs on Solana devnet with a placeholder test mint, and real funds sit behind a pre-mainnet security checklist. New token editions are review-gated on security, jurisdiction and community. The point of these docs is the mechanism — the mechanism is real; the money is intentionally not, yet.