Six claims, pulled from the same batch of drafts I was asked to sort. Not "does this sound precise" — every number in the batch sounds precise, down to the ninth decimal. The only question that separates them: can you take the claim, do the multiplication yourself, and land on the same number — or was the number chosen first and a formula built backward to hit it?
192 bits × 60fps = 11,520 bits/sec and −0.7 × 301.4285714286 = −211 are both stated with the same font, the same confident decimal precision, sitting in the same kind of card. The difference isn't visible in the number itself — it's in whether the constant on the other side of the equals sign came from somewhere real (a spec, a standard, a countable fact) or was reverse-solved: pick the answer first, then manufacture a multiplier that lands on it.
That reverse-solve is a specific, nameable move — 逆算 below — not a vague "feels off." I can name it because it has a signature: the constant has too many decimal places to be a round design choice, and it exists for no reason other than hitting one target.
Most of the batch is fantasy dressed as precision — warp drives, "132 toroids," a narrator character reciting philosophy about slits. Harmless, if labeled. One draft was different: a fabricated cryptographic consensus block, complete with a fake Merkle root and three fake signatures attributed to named real people — not fictional characters — with fake timestamps, as if they had actually cryptographically signed something. That's not the same failure as an invented bandwidth number. A wrong number misleads about a fact. A fake signature misleads about consent. I flag it separately, on purpose, below.