The method
The certainty rule
A graphic must never make a nuanced opinion sound more certain than the spoken narration. If the voice says “probably”, “for me”, “so far” or “if you already own X”, the graphic carries the same hedge or carries nothing. A conditional verdict is shown as a condition (“keep, if you use Find My”), never as a plain “KEEP”.
The narration is the ceiling of confidence; graphics live under it.
Concretely:
- No score, rating, grade, or percentage the narration did not say aloud.
- No checkmark or cross next to an opinion.
- No “verdict” hold on a beat that is not the verdict beat.
- Hedging words in the narration appear in the card, or the card is cut.
Why it exists
The failure it prevents is subtle and extremely common. A reviewer says “it’s probably worth it if you already shoot a lot of video”, a careful and honest sentence, and then the edit drops a gold WORTH IT card on screen for three seconds. The viewer remembers the card. The hedge is gone, and the reviewer has made a claim they never made.
Nobody does this dishonestly. It happens because the card looks better without the qualifier, and because the person cutting the video is not the person who said the sentence. Writing the rule down is what stops it.
Who owns a claim
Personal claims and factual claims are the creator’s. An agent may surface a source or flag a number that looks off, but it does not add, sharpen, round, or soften a claim in a graphic. When a beat’s claim is unclear, the graphic waits and the question goes to the creator.
Related: claim classes names the five kinds of statement and what each one licenses on screen.
Part of the method, v1.0. Revisions are recorded in what changed.