Inversion Tool

Inversion Tool — Munger's Move

"All I want to know is where I'm going to die, so I'll never go there." Charlie Munger's recurring move: instead of asking how to achieve X, ask what would guarantee not-X. Then ensure none of those failure conditions are present. The negative space defines the positive solution more sharply than direct optimization does.

Inversion is distinct from steel-manning (which builds the strongest opposing argument) and from frame-rotator (which produces alternative formulations of the same goal). Inversion takes the same goal and asks: what makes failure certain?

When Inversion Helps

| Helps when | Doesn't help when | |---|---| | Multiple paths to success, hard to compare | Single obviously-correct path | | Failure modes are easier to enumerate than success modes | Pure-creative problems with open-ended success | | Asymmetric outcomes — avoid catastrophe matters more than maximize | Symmetric optimization with clear objective function | | Plan is in pre-mortem stage (before execution) | Plan is in execution — inversion-derived preconditions should already be in place |

The skill is most useful for decisions and plans where the stakes are real and failure has clear shape. Less useful for pure exploration.

Process

Step 1 — Restate the statement as a success criterion

If the statement isn't already framed as "we want X" or "succeed at X", reformulate so the success state is explicit. Inversion needs a clear success-failure axis.

Step 2 — Enumerate failure modes

Ask: "what would guarantee not-X?" Generate the failure catalog. Aim for 5–10 failure modes covering different layers:

| Layer | Example failure modes | |---|---| | Strategic | Wrong target; missing the actual problem; pursuing the wrong metric | | Tactical | Right target, wrong approach; insufficient resources; bad sequencing | | Operational | Right approach, broken execution; missed handoff; quality drift | | Cultural | Misaligned incentives; coordination failure; political resistance | | External | Environmental shifts that invalidate the plan; black swans |

Be specific. "Things go wrong" is not a failure mode. "We launch in Q3 but our competitor pre-empts in July" is.

Step 3 — Negate each failure mode

For each enumerated failure, state the opposite condition — the precondition that must hold for that failure not to occur.

Example:

• Failure mode: "We launch in Q3 but competitor pre-empts in July"
• Negation (precondition): "Either we accelerate to launch by July, OR we differentiate enough that being second doesn't kill the play"

Step 4 — Aggregate preconditions

The set of negated failure modes forms a list of preconditions for success. This is sharper than a generic success spec.

Step 5 — Cross-check with original

Compare the precondition list to the original statement. Are any preconditions missing from the original framing? Those are the candidates the original statement was leaving implicit — surface them.

Output Format

INVERSION — [first 60 chars of original statement...]
─────────────────────────────────────────────
Original goal:   [success criterion restated explicitly]

Failure catalog (what would guarantee not-X):
  Strategic:
    F1. [failure mode] — specificity check: [specific or vague]
  Tactical:
    F2. [failure mode]
  Operational:
    F3. [failure mode]
  Cultural:
    F4. [failure mode]
  External:
    F5. [failure mode]

Negation-derived preconditions:
  P1. [opposite of F1]
  P2. [opposite of F2]
  P3. [opposite of F3]
  P4. [opposite of F4]
  P5. [opposite of F5]

Preconditions NOT in the original framing:
  - [precondition that the original statement leaves implicit; this is the value of inversion]

Inverted statement (alternative formulation):
  [The problem reframed as "ensure none of the failure modes are present" — useful when this
   framing is more tractable than the direct "achieve success" framing]

Distinction from Adjacent Skills

| Skill | What it does | When to prefer | |---|---|---| | inversion-tool | Asks "what guarantees failure"; derives preconditions by negation | Decision/plan with clear failure shape | | steel-man-forge (philosophy/dialectical-tools) | Builds strongest opposing argument | Persuasion or stress-testing a position | | frame-rotator (binding-vow) | Produces alternative formulations of the same goal | Original framing feels narrow | | xy-detector (binding-vow) | Detects solution-disguised-as-problem | Statement is mechanism-shaped |

These can chain. For high-stakes decisions, run frame-rotator + inversion-tool together: frame-rotator opens the solution space; inversion-tool sharpens the success conditions inside that space.

Failure Modes

| Failure | Response | |---|---| | Failure modes are vague ("things go wrong") | Force specificity — name actors, timelines, mechanisms. Vague failure modes produce vague preconditions | | Failure mode is unfalsifiable ("we lack the right culture") | Call assumption-excavator to surface what specifically the cultural-failure framing assumes; convert to a falsifiable mode | | Cannot generate 5+ failure modes | The statement may be too narrow for inversion — escalate to frame-rotator to widen scope first | | Preconditions list duplicates the original success criterion verbatim | Inversion isn't producing new insight; the original framing is already comprehensive. Document this and return |

Output Contract for six-eyes

Called from Phase 4 (Reframe) in standard or deep mode. Returns:

• Failure catalog
• Precondition list
• The "preconditions NOT in original" — this is the load-bearing output for downstream phases
• Optional: inverted-statement alternative formulation

six-eyes uses the "missing preconditions" set to determine if the original statement is missing structurally important elements. If 2+ missing preconditions surface, the statement should be re-stated with those preconditions made explicit before continuing to compression.

Connections

• frame-rotator (binding-vow) — sibling reframing skill; often run together
• steel-man-forge (philosophy/dialectical-tools) — adjacent but different operation
• assumption-excavator (philosophy/logic) — escalation path for vague failure modes
• statement-grader (binding-vow) — downstream consumer; missing-preconditions feed the specificity and scope axes

Sources

• Munger, C. (collected speeches and Poor Charlie's Almanack) — the canonical contemporary articulation of inversion as a thinking tool.
• Jacobi, C. G. J. — "Invert, always invert" — the historical mathematical anchor.
• Heath, C. & Heath, D. (2013). Decisive. — pre-mortem as the structured form of inversion in decision contexts.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: npbuilds
• Source: npbuilds/skill-library
• License: MIT
• Homepage: https://skill-library-prod.web.app/

Install and usage instructions live in the source repository linked above.