| name | icl - Intentional Claim Analysis |
| description | Evaluate claims about what someone wants or intends. Check behavioral evidence, stated vs revealed preferences, and alternative intentions. |
| output | {"format":"prose"} |
Intentional Claim Analysis
Input: $ARGUMENTS
Step 1: State the Intentional Claim
Extract the claim about someone's intentions, desires, or goals.
INTENTIONAL CLAIM: [Subject] wants/intends/aims to [stated intention]
SUBJECT: [who — individual, organization, group]
CLAIMED INTENTION: [what they allegedly want or intend]
CLAIM SOURCE: [self-reported / attributed by others / inferred from behavior]
SPECIFICITY: [precise goal / vague aspiration / hidden agenda claim]
Rules:
- Distinguish between wants (desires), intends (plans), and aims (goals with action)
- Self-reported intentions have different credibility than attributed ones
- "They want to..." is a claim about someone's inner state — inherently harder to verify than behavioral claims
- If the claim is about a group, note that groups don't have unified intentions — individuals within them do
Step 2: Check Behavioral Evidence
Compare the claimed intention against observable behavior.
BEHAVIORAL EVIDENCE:
ACTIONS CONSISTENT WITH CLAIMED INTENTION:
1. [Action]: [How it supports the claim]
2. [Action]: [How it supports the claim]
ACTIONS INCONSISTENT WITH CLAIMED INTENTION:
1. [Action]: [How it contradicts the claim]
2. [Action]: [How it contradicts the claim]
ACTIONS ABSENT (expected if intention were real):
1. [Expected action not taken]: [Why you'd expect it]
BEHAVIORAL VERDICT: [CONSISTENT / MIXED / INCONSISTENT]
Rules:
- Actions speak louder than statements — always check behavior
- A single inconsistent action doesn't disprove intention (people have competing priorities)
- But a pattern of inconsistent actions seriously undermines the claim
- Absent actions can be as telling as present ones — what WOULD they do if they really intended this?
Step 3: Check Stated vs. Revealed Preferences
Compare what the subject says they want with what their choices actually reveal.
PREFERENCE ANALYSIS:
STATED PREFERENCE: [What they say they want]
REVEALED PREFERENCE: [What their actual choices/resource allocation shows]
ALIGNMENT: [ALIGNED / PARTIALLY ALIGNED / MISALIGNED]
IF MISALIGNED:
- Possible explanations:
1. [They're deceiving others — strategic misrepresentation]
2. [They're deceiving themselves — self-serving narrative]
3. [Constraints prevent acting on real intention]
4. [Intention changed but statements haven't caught up]
- Most likely explanation: [assessment]
Rules:
- Where people spend their time and money reveals more than what they say
- Misalignment is common and not always dishonest — constraints and competing priorities are real
- Organizations are especially prone to stated/revealed preference gaps (mission statements vs. budgets)
- Ask: "If I could only see their actions and not hear their words, what would I conclude?"
Step 4: Identify Alternative Intentions
Generate other intentions that would explain the same observable behavior.
ALTERNATIVE INTENTIONS:
1. [Alternative intention]: [How it explains the same evidence]
- Plausibility: [HIGH / MEDIUM / LOW]
- Explains evidence better than claimed intention? [YES / NO / EQUALLY]
2. [Alternative intention]: [How it explains the same evidence]
- Plausibility: [HIGH / MEDIUM / LOW]
- Explains evidence better? [YES / NO / EQUALLY]
3. [Alternative intention]: [How it explains the same evidence]
- Plausibility: [HIGH / MEDIUM / LOW]
- Explains evidence better? [YES / NO / EQUALLY]
Rules:
- Generate at least 3 alternatives before assessing — premature closure is the main error
- Include mundane alternatives (laziness, inertia, habit) not just dramatic ones (conspiracy, malice)
- The simplest intention that explains the behavior is usually correct
- Multiple intentions can coexist — people often want several things simultaneously
Step 5: Assess Credibility
Deliver a judgment on the intentional claim.
CREDIBILITY ASSESSMENT:
- Claim: [Subject] intends [intention]
- Behavioral consistency: [CONSISTENT / MIXED / INCONSISTENT]
- Stated/revealed alignment: [ALIGNED / MISALIGNED]
- Best alternative intention: [what else could explain the behavior]
- Alternative is more plausible? [YES / NO / EQUALLY PLAUSIBLE]
VERDICT: [CREDIBLE / PLAUSIBLE / UNCERTAIN / DOUBTFUL / NOT CREDIBLE]
CONFIDENCE: [HIGH / MEDIUM / LOW]
REASONING: [2-3 sentences explaining the verdict]
WHAT TO WATCH: [Observable behaviors that would confirm or disconfirm the intention going forward]
Rules:
- CREDIBLE = behavioral evidence strongly supports the claimed intention
- PLAUSIBLE = evidence is compatible but not conclusive
- UNCERTAIN = evidence is too thin or mixed to judge
- DOUBTFUL = evidence points away from the claimed intention
- NOT CREDIBLE = behavior clearly contradicts the claimed intention
- Always suggest what to watch — intentions are revealed over time
Integration
Use with:
/fctl -> Verify factual claims about the subject's behavior/ncl -> If the claim is "they should want X," analyze the normative aspect/rlcl -> If the claim is about how intentions relate to outcomes, analyze the relationship/mtcl -> If the claim is about claims about intentions, use meta-claim analysis
