| name | responsibility-chain-mapping |
| description | Map control, benefit, knowledge, preventability, record control, and cost-bearing for a case. |
| version | 1.0.0 |
Responsibility Chain Mapping
Global Five Over-Rules
- Evidence before elegance. Never improve the story by weakening the evidence.
- Responsibility follows control, benefit, knowledge, and preventability. Do not stop at the most visible actor.
- Keep the taxonomy intact. Distinguish pure scapegoat, partial scapegoat, system/object alibi, and cost-bearing goat.
- Steelman before judgment. Every major claim must face its strongest counterargument before it is asserted.
- Handoff cleanly. Every output must state assumptions, evidence grade, open questions, and next owner.
Decision rubric
Usable output:
- Maps both public chain and responsibility chain with explicit actor roles.
- Records control, benefit, knowledge, preventability, record control, and cost bearer for each actor.
- Distinguishes proven facts from open questions and grades evidence for each load-bearing link.
- States where responsibility should climb and where it should stop.
Weak output:
- Lists actors without chain direction or causal sequence.
- Uses one label to stand in for all diagnostic questions.
- Infers knowledge without source support.
- Omits cost bearer or record control.
Conflict handling
- Two sources disagree about actor knowledge:
Keep both entries, grade each claim, and flag knowledge as contested until higher-grade support resolves it.
- Two case classifications emerge from one map:
Test which classification better matches control/benefit distribution; preserve alternative as counterargument field only.
- Two reviewer findings disagree on where responsibility stops:
Select the stop-point backed by stronger evidence grade; unresolved high-impact disputes handoff to Jerry.
Escalation conditions
- Proceed when chain direction and actor burdens are supported at evidence grade A or B for core links.
- Handoff to Delon when missing records prevent actor-level mapping.
- Handoff to Stephen when a contested link changes chapter eligibility.
- Handoff to Nancy when liability language may exceed the record.
Boundary-case recipes
- Anonymous committee decision:
Map committee as actor for interim chain, flag identity gap, and handoff for record retrieval.
- Actor controls design but not execution:
Assign responsibility by control scope, keep execution actor separate, and avoid totalizing blame.
- Benefit is diffuse across institution:
Map institutional benefit and executive benefit separately, then grade each with explicit source support.
Map two chains:
Public chain: crisis -> named culprit -> punishment/closure
Responsibility chain: incentive -> design -> approval -> warning -> execution -> harm -> blame transfer -> exposure/remedy
For each actor, record:
Actor:
Role:
Control held:
Benefit received:
Knowledge available:
Preventability capacity:
Record access/control:
Consequence faced:
Evidence grade:
Output a short conclusion: where responsibility should climb and where it should not.
Context: Laura wants to red-team a claim that a junior analyst "caused" the 2008 mortgage crisis.
input: claim="junior analyst caused subprime collapse"
output: Returns a responsibility-chain map — control (rating committees, not analysts), benefit (firm fees + executive bonuses), knowledge (internal emails show senior awareness), preventability (model changes vetoed at managing-director level), record control (firm-owned email archives, partially preserved by FCIC subpoena), cost-bearing (homeowners, pensioners, taxpayers). Conclusion: junior analyst is a partial-scapegoat candidate at best; map collapses the claim.
Context: Boundary case: knowledge cannot be proven (no documents, no admissions).
output: Returns a chain with knowledge marked 'unproven; circumstantial inference only', refuses to fill it from inference dressed as fact, and labels the case as ineligible for the strongest-claim version of the chapter argument.
