The Forge Adjudicator Card
Two members have written things that cannot both be true. This is normal, it is not a failure, and it is the single most common moment a shared world stalls — because the argument on the table is whose idea was better when the question that actually has an answer is which one the world can carry. The card walks that second question in four steps and writes the ruling down.
Step one — cite both entries
in the builders’ own wordsWrite each entry as its builder would recognise it, and name them. A ruling made against a paraphrase is a ruling the builder will not accept, because the version in the card is not the version they wrote.
Step two — check each against the canon spine
what the world already carriesThe spine is whatever the world has already committed to elsewhere: earlier rulings, published entries, the geography everyone has been writing against. An entry that contradicts the spine is not wrong — it is expensive, and the cost lands on everyone who has already built on the old fact. Say which spine elements each entry touches, and say how many existing entries would have to change if it won.
Step three — choose the ruling
adopt · blend · retireThree verdicts, and no fourth. Adopt one and retire the other. Blend them into a single entry that keeps what each builder actually cared about. Retire both and write the thing they were both failing to say — rare, but sometimes the collision has revealed that the real fact was neither. Circle one and say why in a sentence the builders can repeat to someone else.
Step four — record it and credit both builders
the line that goes into canonWrite the final entry exactly as it will appear, then name every builder whose work survives into it — including, and especially, the one whose version lost. A contributor who is credited for the part of their idea that did last will bring another one.
Copy-paste wording for each verdict
so the ruling lands the same way twiceRulings read badly when they are improvised. These are the three phrasings, ready to paste into the shared-world page beside the recorded entry.
Adopt. “We are keeping [entry], credited to [builder]. The other version is retired — not because it was worse, but because the world can only carry one, and this one costs fewer existing entries.”
Blend. “We have kept what both entries were reaching for in one line: [entry]. Credited to [both builders]. The collision was useful — neither original version had [the part that only works together].”
Retire. “Neither entry stands. What both were circling is [entry], which neither had said yet. Credited to [both builders] for finding it by colliding.”
A collision is a gift that arrives looking like a problem
Two members writing incompatible facts about the same place means two people care about the same corner of the world — which is exactly what a shared world is for. The failure is not the collision. It is treating it as a contest of whose idea is better, which has no answer and leaves a loser, rather than which the world can carry, which does.
That is why step two asks about the spine rather than about quality. A beautiful entry that invalidates three existing ones is not disqualified — it simply has a price, and the circle should be the one deciding whether to pay it, with the number in front of them instead of discovering it afterwards. Writing the price down turns an argument about taste into a decision about cost.
Record every ruling, including the ones that felt obvious. The ruling is what lets someone who was not at the meeting build on the result without re-litigating it — and crediting the losing builder is what keeps them building. A world where losing a collision costs you your name is a world where nobody proposes anything again.