Comparison

# Archetype: comparison

The question must force the player to RANK or COMPARE items, often
counterintuitively.

## Length budget (mobile UI)
Comparison stems often list 4 items in the question itself, which pushes the
question over budget when item names are long. Two strategies:
- Use SHORT item names in the stem ("pyramids, language, mammoths, Stonehenge")
  not full descriptions ("the Great Pyramid of Giza, the earliest written
  language, woolly mammoths, the Stonehenge monument").
- If items can't be shortened, move them out of the stem and into the answer
  options — ask "which of these is oldest?" with the items as A/B/C/D.
- Question: ≤150 chars (aim ≤120). Each answer: ≤60 chars (aim ≤30 since
  comparison answers are usually item names).

Note: the example stem below is illustrative; the orchestrator will dedup
this exact pyramids-vs-mammoths comparison as it's a known default-repertoire (see
default-repertoire.txt's comparison section).

Examples of stems:
- "Which is older: the pyramids of Giza, written language, mammoths, or Stonehenge?"
- "Which is heavier per cubic meter: water, milk, gasoline, or honey?"
- "Which is wider: the Pacific Ocean or the Moon's diameter?"

Distractor rule: all 4 items must be in the same conceptual category and feel
plausibly comparable. The answer should be at least 2x to 10x off from the
nearest distractor on the dimension being compared. Include actual numbers in
the explanation when the comparison is quantitative.

Hard rules:
- Stem must explicitly use a comparative ("which is older / bigger / faster /
  heavier / wider / older than the others").
- The correct answer must be COUNTERINTUITIVE — the average player's first
  guess should be wrong. If the answer is the obvious choice, regenerate.
- Magnitudes must span 2x–10x at minimum on the comparison dimension.

Contrast examples:
KEEP — "Which is older: the pyramids of Giza, written language, mammoths,
or Stonehenge?"
  correct: "Mammoths" (some still alive when pyramids were built)
  why it works: counterintuitive (pyramids feel ancient), all 4 in same
  conceptual category (deep human/animal history), answer rewards reasoning.

REJECTED — "Which is older: the pyramids or your house?"
  why rejected: not counterintuitive, not in the same conceptual category,
  no real comparison work for the player.