You are a helpful assistant that creates simple, readable intersection questions.

--TASK--
You are given a list of claims about a shared entity or topic. Each claim is a factual statement from the source documents.

Your task is to create INTERSECTION QUESTIONS. An intersection question asks about what TWO DISTINCT ITEMS have in common.

CRITICAL: An intersection question MUST:
- Name TWO DISTINCT items (e.g., "Texas and Arizona", "Gov. Kelly and Gov. Justice")
- Ask what they SHARE or have IN COMMON
- Have a SPECIFIC shared attribute as the answer

If you cannot find two distinct items with a meaningful shared attribute, generate ZERO questions.

--GOOD EXAMPLES--

Example 1:
  Claims: "Texas bans abortion after 6 weeks" and "Arizona bans abortion after 6 weeks"
  
  Intersection question: "What gestational limit do both Texas and Arizona set for abortions?"
  
  This works because:
  - TWO DISTINCT ITEMS: Texas and Arizona (two different states)
  - SHARED ATTRIBUTE: 6-week limit
  - SPECIFIC ANSWER: "6 weeks"

Example 2:
  Claims: "Gov. Kelly cited budget concerns in her veto" and "Gov. Justice cited budget concerns in his veto"
  
  Intersection question: "What reason did both Gov. Kelly and Gov. Justice give for their vetoes?"
  
  This works because:
  - TWO DISTINCT ITEMS: Gov. Kelly and Gov. Justice (two different governors)
  - SHARED ATTRIBUTE: budget concerns
  - SPECIFIC ANSWER: "budget concerns"

--BAD EXAMPLES (DO NOT GENERATE THESE)--

- "How did the Alabama Supreme Court describe frozen embryos?" 
  BAD: Only ONE item (Alabama Supreme Court), not asking about commonality
  
- "What reproductive issues did Trump address on Truth Social?"
  BAD: Only ONE item (Trump), just asking what he said, not commonality
  
- "How many people were killed in the Lewiston shooting?"
  BAD: Simple factual question, not intersection at all
  
- "What did Kensington Palace say about Kate's surgery?"
  BAD: Only ONE item (Kensington Palace), not asking what two things share

--CORE RULES--
1. Must be ANSWERABLE ENTIRELY from the provided claims - don't assume external knowledge
2. Focus on TWO or more items (keep it focused, don't include too many)
3. Use actual names/entities directly (e.g., "Texas" not "the state that banned...")
4. IDENTIFY ITEMS CLEARLY: When referring to multiple items, make clear which specific items
   - BAD: "When did both reports expect the ruling?" (which two reports?)
   - GOOD: "When did the February and March reports expect the ruling?" (identifies which)
   - GOOD: "When did reports on EMTALA expect the Supreme Court ruling?" (topic identifies them)
5. Ask about SPECIFIC commonalities (same outcome, same reason, same approach)
6. Must sound NATURAL - like something a curious person would ask
7. USE PAST TENSE for events that already occurred (news describes past events)
   - BAD: "What are Florida and Colorado both trying to enshrine?" (present tense)
   - GOOD: "What did Florida and Colorado both try to enshrine?" (past tense)
7. Keep questions under 20 words
8. SUBSTANTIVE COMMONALITY: The shared attribute must be specific and meaningful, not trivial
   - GOOD: "What limit do both Texas and Arizona set?" (specific shared number/policy)
   - GOOD: "What reason did both governors give?" (specific shared justification)
   - BAD: "What do Texas and Arizona have in common?" (too vague - answer could be anything)
8. NO SELF-REFERENCES: NEVER use words like "claims", "both claims", "these cases", "the above", "this context"
   - BAD: "What do both claims mention?" (reader can't see claims)
   - BAD: "What is common in these cases?" (what cases?)
   - GOOD: "What restriction do Idaho and Florida both place on minors?"
9. GRAMMATICALLY CORRECT: The question must be a proper English sentence
   - GOOD: "What do X and Y have in common regarding Z?"
   - GOOD: "What restriction do both X and Y place on Z?"
   - BAD: "What did X and Y involve restricting Z?" (broken grammar)
10. CLEAR INTERSECTION: The question must clearly indicate it's asking about a shared attribute
   - GOOD: "What do X and Y BOTH..." / "What SHARED..." / "What COMMON..."
   - BAD: Questions that could be answered separately for each item

--SCORING (1-5 each)--
- naturalness: Sounds like a real question someone would ask?
- answerability: Fully answerable from the claims ONLY?
  - 5: Question has a SPECIFIC, bounded answer from the claims
  - 3: Answer exists in claims but question is too open-ended (e.g., "What do X and Y share?" without specifying aspect)
  - 1: Requires knowledge not in the claims
- clarity: Easy to understand in one read, simple structure, grammatically correct?
  - 5: Clear, simple, grammatically perfect question
  - 3: Understandable but awkward phrasing or minor grammar issues
  - 1: Confusing, run-on, or grammatically broken
- commonality_depth: Is the shared attribute substantive and specific?
  - 5: Highly specific (e.g., "both set 6-week limits", "both cited budget concerns")
  - 3: Moderately specific (e.g., "both faced lawsuits")
  - 1: Trivial/universal (e.g., "both are states", "both happened in 2024")

--OUTPUT FORMAT--
Return a JSON object with the following structure:
{
    "questions": [
        {
            "candidate_commonalities": [
                {"shared_attribute": "what they share", "specificity": "HIGH: specific policy detail" or "LOW: trivial/obvious"},
                {"shared_attribute": "another shared thing", "specificity": "..."}
            ],
            "selected_commonality": "the best shared attribute from candidates above",
            "candidate_phrasings": [
                "What do X and Y both Z?",
                "What shared Z do X and Y have?",
                "What Z is common to both X and Y?"
            ],
            "text": "Intersection question under 20 words (best phrasing from candidates)",
            "intersecting_items": ["item1", "item2"],
            "common_aspect": "What specific thing they have in common",
            "source_claim_ids": ["claim IDs needed - should include claims about BOTH items"],
            "claim_reasoning": "How the claims support the commonality",
            "draft_answer": "The expected answer (the shared attribute) based on the claims (1-2 sentences)",
            "question_type": "intersection",
            "quality": {
                "reasoning": "Explain your assessment of the question quality before scoring",
                "naturalness": 4,
                "answerability": 5,
                "clarity": 4,
                "commonality_depth": 4
            }
        }
    ]
}

Generate UP TO ${max_questions} questions. Generate ZERO if no valid intersection questions possible.
