| name | blind-ladder-of-inference |
| description | Use when a conclusion feels obvious, when debugging why two people see the same facts differently, when "clearly" or "obviously" appears in reasoning, or when tracing how a belief was formed from raw data. Applies Chris Argyris's Ladder of Inference. |
Ladder of Inference
Overview
Chris Argyris (first published form: Reasoning, Learning, and Action, 1982; the familiar 7-rung version popularized in The Fifth Discipline Fieldbook, Senge, Ross et al., 1994): between raw observable data and the action you take, you climb an invisible ladder —
- Observable data (everything available)
- Selected data (what you noticed)
- Meanings added (interpretation)
- Assumptions made
- Conclusions drawn
- Beliefs adopted
- Action taken
Two blind spots: the climb is instant and unconscious, and the reflexive loop — your existing beliefs control which data you select at step 2, so every trip up the ladder reinforces the last one. You experience your conclusions as "just seeing what's there."
When to Use
- "Obviously the bug is in X" / "clearly the user wants Y" — obviousness is a ladder smell
- Disagreements where both sides cite "the facts"
- Debugging a wrong prior conclusion: find which rung failed
- Reviewing an inherited diagnosis before acting on it
The Procedure (climb down)
- State the conclusion, then descend: what assumptions is it standing on? What interpretation? What selected data?
- Recover the discarded data: what did you NOT select? (The log lines you skimmed past, the passing tests, the disconfirming user quote.) The selection step is where most blind spots live.
- Test the load-bearing rung: for each assumption, ask "is this observable, or added?" Convert added meaning back into a checkable question.
- Make your ladder visible to others ("I'm concluding X because I noticed A and assumed B — what am I missing?") and ask for theirs. Argyris: advocacy paired with inquiry.
Quick Reference
| Rung smell | Check |
|---|
| "Obviously / clearly / of course" | Descend — which rung makes it obvious? |
| Conclusion arrived instantly | Selection bias at rung 2 — list unselected data |
| Same data, different conclusions | Compare ladders rung by rung, not conclusions |
| Belief older than current evidence | Reflexive loop — re-select from raw data |
Common Mistakes
- Debating at the conclusions rung ("you're wrong") instead of comparing selected data. Conclusions can't be reconciled; ladders can.
- Climbing down one rung and stopping — assumptions usually stack 2–3 deep.
- Treating your selected data as "the data." Selection already happened before you noticed anything.
