How fairness works here

How a verifiable draw works

A verifiable draw fixes the order of a bracket with randomness that nobody involved could choose. The tournament commits first — to its inputs and to a round of a public randomness beacon that has not been published yet — and that round then decides the order.

1. Before the draw, the tournament commits to its inputs — the entrants and the exact procedure — and to a future round of a public randomness beacon.
2. That round has not been published yet, so at that moment its randomness is unknown to everyone — to the organizer and to Arena as much as to you.
3. When the round is published, its value decides the order deterministically. No person picks it, and replaying the reveal does not change it.
4. Anyone can derive that order again from the same published round and the same committed inputs, and compare it with the bracket that was published.

What this proves is narrow and specific: the randomness that set the order was published after the commitment, so it could not be predicted or chosen. It says nothing about how the matches were then officiated, or about how the results were entered.

What the open verifier re-checks

The verifier runs in your browser, over the evidence published for one exact Arena revision, and re-checks the following independently of Arena:

• that the recorded event log replays to exactly the final engine state that is claimed, so the state on the page is the one that log produces and not a separate summary written alongside it;
• that the definition hash and the engine state hash match the identity published with the record;
• that each draw transcript is internally consistent and yields the order that was published;
• that each bound beacon round carries a valid signature from its public source, so the randomness is the real published round and not a substitute.

A verification report declares its own scope. The draw, beacon and notary checks run only for a tournament that claims them; a check outside the scope did not run, and the report makes no statement about it. When required evidence is missing, withheld or unsupported, the result is reported as incomplete — never as verified.

Honest boundaries

Each claim on the record is a narrow statement about specific evidence. The boundaries below are as much a part of a claim as the claim itself.

A seal does not mean that cheating is impossible. A seal says that a commitment over the published state exists. It does not say that the checks passed, that the officiating was sound, or that the results people entered were honest. Arena can show that a record was not altered after the fact; it cannot show that the record was true when it was written.

Manual seeding carries no unbiasability claim. When the order is set by a person or imported from elsewhere, the record says exactly that: the order is manual, and nothing is claimed about it being unpredictable. A manual order can still be replayed and checked, but that shows only that it was applied unchanged.

Manual recovery is a visible downgrade, not a repair. If a verifiable draw fails or expires and the organizer finishes it by hand, the failed attempt stays in the record, the order becomes manual at a new Arena revision, and the unbiasability claim is dropped rather than carried over.

A reproducible replay is not an unbiasability claim. Replay shows that the committed event log leads to the state on screen. Whether the order behind that state was unpredictable is a separate question, and only a valid beacon binding answers it.

A field outside the supported size falls back to manual seeding. The ceremony works over a bounded number of entrants. Outside that range the tournament is seeded by hand rather than blocked, and the record shows the manual order in the usual way.

Every claim is bound to one exact revision. A verification result names the snapshot, the Arena revision, the engine event revision, the engine implementation version and the engine state hash it checked. A later revision is not covered by it and needs a result of its own.

The gap between commitment and publication

A public beacon means the randomness could not be predicted. On its own it does not prove that only one commitment was made. Without an independent witness, the moment of commitment relative to the beacon round is Arena’s own record, and an organizer could in principle run the ceremony privately more than once and publish only the outcome that suited them.

An independent notary closes that gap for a draw. The notary countersigns the single commitment before the beacon round is published, so the ordering no longer rests on Arena’s word alone. The countersignature is checked against a key pinned in the released verifier; a key the verifier does not recognise produces no notarization claim at all, rather than a weaker one.

Whether a draw was notarized is shown on that draw, not promised here. Notarization applies to one draw at a time. A draw carries a notarized claim only when a recognised countersignature was checked and passed; otherwise it states plainly that it is not notarized, and the gap described above applies to it.

Check it yourself

Take none of this on trust. Open the Verify page of a tournament to run the checks in your own browser, or run the open verifier yourself over the evidence published for the exact revision you care about.