Supported well enough for this decision. Still open to correction if better evidence appears.
Method
Find the gap before you choose a lens.
Two parts. The Reality Map shows where to look and what kind of statement you are actually holding. The lenses supply the question once the gap is visible.
How to use it
You do not fill the map. You scan it for the gap.
Most of the map stays empty, and that is correct. The point is to notice the one missing statement type or layer that would actually change the decision — not to produce a complete picture.
The Reality Map is the structure of a First Position: it is how you separate what you know from what you have already explained, before anyone else speaks.
A decision can fail because the wrong layer was modelled, or because an observation, an interpretation and an assumption were all treated as the same kind of statement. The map checks both.
Axis 1 · Epistemic type
First label what kind of statement you are dealing with.
Who saw it, when, in what context. One observation is a signal, not automatically a general law.
An explanation laid on top of what happened. Useful — and not the same thing as what happened.
A dependency inside the model that has not yet earned fact status.
Something decision-relevant that we do not currently know. Usually where the gap is.
Worked example
One question, sorted into five kinds of statement.
Original question: should the commercial director be fired?
The role, the contract, the targets and the reporting line are documented.
Three team members stopped escalating to him during the last quarter.
“The team lost trust in him.”
Lost trust is the main cause of the missed target.
What direct reports say privately, and which decisions are genuinely blocked rather than slow.
The Second Question
Which constraint keeps making strong leaders fail in this role?The original question kept the decision inside the map that produced the dead end. Nothing here proves the director should stay. It moves the next step from a verdict to a test.
Illustrative composite, not a participant case.
Axis 2 · Reality layer
Then ask where the mechanism actually lives.
The two axes together
A statement about people can still be an assumption.
The axes are independent. Keeping them apart is what stops an explanation from being filed as a fact.
| Layer | Fact | Observation | Interpretation | Assumption | Unknown |
|---|---|---|---|---|---|
| People & incentives | Role, contract, KPI | Three reports stopped escalating | The team lost trust | Trust is the main cause of the miss | What reports say privately |
| Systems & processes | Decision rights are documented | Approvals keep returning to the founder | The director lacks authority | More authority would raise throughput | Which decisions are actually blocked |
| Environment | New regulation takes effect on date X | Competitors changed pricing | Market pressure is rising | Pressure will hold for 12 months | How elastic demand really is |
Scroll the table sideways to see every column.
The highlighted column is the point of the exercise. In most stuck decisions the useful move is not a better argument in the columns you have already filled — it is one cheap way to convert an unknown.
Provenance
Consensus is weaker when everyone relies on the same source.
Five similar estimates built on one report are not five independent pieces of evidence.
So the map also asks where a signal came from: direct observation, personal experience, public evidence, another participant, a second-hand report, or intuition.
This is why a First Position is written before discussion, and why an AI answer is treated as one more correlated opinion rather than a second judgement.
It is also the reason the room is worth more than one good advisor: independence is the thing being protected.
Lenses
Once the gap is visible, pick the question that fits it.
A lens is not the correct map of the world. It is a question that helps reveal a typical error, and every question has limits.
empirical componentsWhat is observation here, and what is already a causal story?Fact and explanation · Cognitive bias · Kahneman
Useful when: When confidence rests more on wording than on data.
Limit: The boundary is not always clean: observation also depends on measurement and context.
Evidence status: empirical components.
heuristicIs this a constraint or a symptom of a deeper constraint?Constraint and root cause · Theory of Constraints · Goldratt
Useful when: When the same problem returns after local fixes.
Limit: Searching for one root cause can oversimplify a system with interacting causes.
Evidence status: heuristic.
heuristicWhich options disappeared because we silently treated something as fixed?Breaking a hidden assumption · Lateral thinking · de Bono
Useful when: When the decision is trapped between two familiar options.
Limit: Generating unusual options does not establish feasibility or value.
Evidence status: heuristic.
empirical methodWhat usually happens in a genuinely comparable class of situations?Outside view · Reference classes · forecasting
Useful when: When an inside story of uniqueness displaces evidence from similar cases.
Limit: A poor reference class creates false precision.
Evidence status: empirical method.
empirical components + heuristicIf the decision failed, which mechanism most plausibly broke it?Inversion and premortem · Inversion · premortem
Useful when: Before action with meaningful downside or a complex execution chain.
Limit: It can become a list of fears without ranking or testing.
Evidence status: empirical components + heuristic.
heuristic / philosophicalHow can we preserve room to manoeuvre and limit irreversible harm?Optionality and ruin · Optionality · Taleb · inversion
Useful when: When uncertainty is high and a full commitment creates unrecoverable downside.
Limit: Excessive caution also has a cost and may block learning.
Evidence status: heuristic / philosophical.
heuristic / contestedAre we disputing facts, or using different criteria for a good outcome?Stakeholder and worldview · Systems · stakeholder views
Useful when: When parties see the same data but judge outcomes through different values and incentives.
Limit: A philosophical map must not hide power asymmetry or incompatible interests.
Evidence status: heuristic / contested.
experience-basedWho has seen this from the inside, and what do they know that public data does not?Tacit knowledge · Independent participants
Useful when: When execution practice, local context, or informal constraints matter.
Limit: One person’s experience may be unrepresentative and still needs testing.
Evidence status: experience-based.
An author’s authority does not replace evidence. Empirical, heuristic, philosophical and contested are not prestige ratings; they indicate how cautiously a lens should be used.
The Reality Skill practice
Components have different support. The combination remains a working hypothesis.
First Position and The Return create a measurement frame, but participant value should arise in expanding the map.
Decision records reveal patterns first. The term base rate is reserved for a sufficiently homogeneous class with a clear denominator.
After each outcome, the record includes not only participant error but what the Chartroom or the practice missed.
Next
Run the map on a decision you actually carry.
The map is worth little in the abstract and quick on a real question with a real deadline.