human-avatar Investigation: Claim Decomposition
Complex claims are bundles. "The talent market has fundamentally shifted" contains claims about direction, magnitude, durability, cause, scope, and relevance — each independently falsifiable, each carrying different amounts of load. When people debate such a claim, they are usually debating different sub-claims without realizing it. Decomposition makes the hidden architecture visible: what is being asserted, what is being assumed, what is verifiable, and what is doing the actual load-bearing work.
Your Process
Step 1: State the Master Claim Write out the claim in its current form. If it was stated imprecisely, note that — but trace it faithfully. Do not pre-interpret; decompose what was actually asserted.
Framing check: Confirm the specific claim before continuing. State what you've identified — the actual claim being decomposed and its apparent domain (factual, causal, normative, etc.) — in one sentence, then use AskUserQuestion:
- Question: "I'm reading this as: [your one-sentence framing of the claim and its domain]. Is that right?"
- Header: "Framing"
- Options:
- Yes — proceed — framing is correct
- Adjust — one element is off; user will correct it before you continue
- Reframe — different claim than read; incorporate the correction before proceeding
Step 2: Decompose into Atomic Sub-Claims Break the master claim into the smallest independently assessable statements. Each sub-claim should be:
- A single, specific assertion
- Independently falsifiable (you could in principle evaluate it separately from the others)
- Not a combination of two ideas that are themselves separable
Common decomposition dimensions:
- Factual claims — assertions about what is true right now
- Causal claims — assertions about what causes what
- Trend claims — assertions about direction, rate, or duration of change
- Comparative claims — assertions about relative magnitude (more than, better than)
- Definitional claims — assertions about what a term means or what counts as X
- Normative claims — assertions about what should be done or what is good
- Scope claims — how broadly the main claim applies (always, usually, in some contexts)
Aim for completeness: a claim with 3 surface sub-claims usually has 6-10 once decomposed carefully.
Step 3: Classify Each Sub-Claim For each sub-claim, assign a classification:
| Classification | Meaning |
|---|---|
| Verified | Well-established, supported by strong independent evidence |
| Checkable | Testable in principle; evidence exists or could be gathered |
| Contested | Active disagreement in the relevant domain; competing evidence |
| Uncheckable | Unfalsifiable by design (definitional, philosophical, or unprovable) |
| Assumed | Treated as true without being argued; load-bearing but unexamined |
Step 4: Identify the Load-Bearing Sub-Claims
Before narrowing: Show the complete list of classified sub-claims to the user first. Use AskUserQuestion:
- Question: "I've identified [N] sub-claims. Before I select the most load-bearing ones, are there any you'd flag as especially critical, or any I've missed?"
- Header: "Prioritise"
- Options:
- Proceed with your selection — the set looks right
- Flag one — user will name a specific sub-claim to include as load-bearing
- Add a missing one — user will describe a sub-claim not yet listed
Not all sub-claims carry equal weight. Determine:
- Which sub-claims, if false, would cause the master claim to fail?
- Which sub-claims are doing heavy lifting that isn't acknowledged?
- Which assumed sub-claims are most at risk of being wrong?
These are the critical nodes: the sub-claims where investigation effort should concentrate.
Step 5: Map the Dependency Structure Do some sub-claims depend on others being true first? Build a simple dependency map: "Sub-claim C requires sub-claim A and B to hold." This reveals whether the whole claim stands or falls on a single contested premise.
Step 6: Recommend Verification Priorities Given the load map and classifications, what should be verified first? Rank by:
- How much load it carries (does the master claim fail if this is false?)
- How checkable it is (is verification feasible?)
- How likely it is to be wrong (is this an assumption that has been examined?)
Human Check-in
Before proceeding, use the AskUserQuestion tool. State your interpretation of the situation in 1–2 sentences — what is being analyzed and what the core question is — then ask:
- Question: "My read: [your 1–2 sentence interpretation]. How do you want to proceed?"
- Header: "Scope"
- Options:
- Full decomposition — Complete all steps: decompose, classify, map dependencies, recommend priorities
- Decompose and classify only — List all sub-claims with classifications; skip dependency mapping
- Load-bearing claims only — Focus on identifying which sub-claims are doing the most work
- Refine the claim first — Clarify what exactly we're decomposing before starting
Proceed based on their selection. If the user reframes, incorporate the correction before running any analysis.
Output Format
Master Claim
[The claim as stated]
Sub-Claim Decomposition
| # | Sub-Claim | Type | Classification | Notes |
|---|---|---|---|---|
| 1 | [sub-claim] | Factual / Causal / Trend / Comparative / Definitional / Normative / Scope | Verified / Checkable / Contested / Uncheckable / Assumed | [Brief note] |
| 2 | ... |
Load-Bearing Sub-Claims
Highest load:
- [Sub-claim #] — [Why it's load-bearing: what fails if it's false]
Critical assumptions:
- [Sub-claim #] — [What is being taken as given; what would have to be true]
Dependency Structure
- [Sub-claim C] depends on [Sub-claim A] and [Sub-claim B]
- [The whole claim collapses if [sub-claim X] is false]
Verification Priorities
- [Sub-claim #] — High load, checkable, hasn't been examined: [how to verify]
- [Sub-claim #] — Assumed; likely the weakest point: [what would falsify it]
- [Sub-claim #] — Contested in domain; worth checking current state: [what the debate is]
Notes
Use investigation-evidence-audit once you've identified the load-bearing sub-claims and want to evaluate the evidence for those specific parts. Use investigation-source-trace when you want to know who first made the master claim and how it has changed. Claim decomposition is specifically about the internal architecture of a claim — not whether its evidence is good, but what it is actually asserting.
What's Next
After delivering this output, use AskUserQuestion to offer the next move:
- Question: "Claims decomposed. What's next?"
- Header: "Next"
- Options:
/s4h-investigation-source-trace— Trace each component claim to its source/s4h-investigation-evidence-audit— Audit evidence for each component claim/s4h-logic-check— Validate the inference structure between claims- Done — Wrap up and synthesise what we have so far