What the commitment hash does not prove

The hash proves the pick existed in this exact form before kick-off and was not edited afterwards. It does not prove the pick was good. And — the part most sites in this business would leave out — it does not prove when we first published it.

It does not prove when we published it

The commitment is self-attested. We publish the hash on our own server, timed by our own clock. From out here, a server quietly re-signing a different hash would look identical to a server that has been showing you the same one all along. So what the hash rules out is a row being edited after you have seen it. On its own, it does not rule us out.

What would actually settle it is a signed time from somebody who is not us: an RFC 3161 timestamp, where the authority only ever sees the hash and never the pick, with an OpenTimestamps Bitcoin anchor on top. Even that would not be enough alone — a stamp is only worth something beside a complete, gapless index, or we could stamp ten picks and show you the one that landed. The index exists, one row per settled day, nothing removable. The stamping does not. Until it does, this page will keep saying so, and so will the results archive, which prints the same limitation under a heading that reads Unverified.

It does not prove the pick was good

A committed pick is a committed opinion. Freezing it in advance makes the opinion checkable; it does not make it right, and it does not take the chance out of a game whose result is not ours to control. A pick committed six hours before kick-off that lost is exactly as committed as one that won. If a commitment scheme could tell you which was which in advance, it would not be a commitment scheme, it would be a crystal ball.

What the hashed string does not cover

• The side. The seven hashed fields are the date, the fixture, the market, the two prices, the kick-off and the salt. The side is recorded on the pick's own row, but it is not one of the seven — so what the hash pins is which game, which market and at what price, not which way we went.
• Anything written at settlement. The final score, the verdict and the settlement time are added after the commitment was made, so the hash says nothing about them. A page that quietly edited those would not announce it.
• The reading of the result. The settlement rules name the source and the tiebreak and are served as data rather than typed onto a page — but today a person on our side looks up the public result and marks the row. The rule is declared; the reading is still ours.

None of this is a reason to trust the archive less than you would trust anybody's word. It is a reason to know exactly which parts rest on our word: today, the timing and the reading. Everything else you can check yourself.