Choose a tool when it helps answer a concrete question. The doctrine governs the commitment; a thinking tool helps investigate it. The tools below were selected from Untools for their fit to story design and change. The examples and diagrams here are original applications.
| Your Question | Useful Tool | Expected Result |
|---|---|---|
| Are we solving the right problem? | Abstraction laddering | A clearer purpose and a bounded problem |
| Which concepts are being confused? | Concept map | Explicit relationships that can be read as statements |
| What could account for this failure? | Issue tree | Distinct investigation branches within a named scope |
| Are we treating an assumption as a fact? | Ladder of inference | A separation of observation, interpretation and conclusion |
| What else changes if we do this? | Second-order thinking | Consequences for consumers and outstanding work |
| How could this promise be broken? | Inversion | Specific counterexamples to investigate or prevent |
Abstraction Laddering
Use this when the request already assumes a solution. Move toward purpose with “why?” and toward a concrete response with “how?”. Untools: abstraction laddering.
Example: “Add an approved flag” assumes a representation. The purpose is to
publish only the reviewed version. Possible mechanisms now include a version-bound
decision, a protected prepared value, or a stored comparison at publication.
The upward question does not authorize a wider product scope. The downward answer still needs to preserve the owner's actual requirement. This tool frames a question; it is not Zachman's transformation axis or a proof of the answer.
Concept Map
Use this when people use the same words for different things. Name the relevant concepts and connect them with explicit relationships. Untools: concept map.
Read each edge aloud: “an occurrence binds actual resources.” If the sentence cannot be defended, the map has exposed a missing definition. A drawn edge is not an implemented relationship; follow it into the owning code or evidence.
Issue Tree
Use this to divide an investigation into testable branches. Keep the chosen scope explicit and check that branches neither overlap nor omit an intended case. Untools: issue trees.
For a report that “the wrong announcement version was published,” begin with:
| Branch | Evidence to Inspect |
|---|---|
| No decision record matches the reported published version. | Decision lookup, version binding and the actual publication record |
| A decision record matches the reported published version. | Its issuer, content binding, current validity, and whether the report describes the actual publication |
This first split classifies the incident; it does not identify the cause yet. Continue only where observations justify it. Do not call a convenient checklist exhaustive merely because it has tidy headings.
Ladder of Inference
Use this before turning an observation into a decision. Separate what was seen from its selected meaning and the assumptions used to reach a conclusion. Untools: ladder of inference.
| Step | Timeout Example |
|---|---|
| Observation | No acknowledgement arrived before the deadline. |
| Interpretation | The caller lacks a result. |
| Unsupported assumption | The source did not apply the effect. |
| Unsafe conclusion | Sending it again is harmless. |
| Better next evidence | Reconcile the original operation with the source. |
A typed error named Timeout does not validate the assumption. This lens is
especially useful when a model, UI or log message claims more than its observation.
Second-Order Thinking
Use this when a locally useful change can affect another promise or later event. Follow consequences beyond the immediate benefit. Untools: second-order thinking.
Example: route new messages through provider B. New work can use B, but an unknown operation already sent to A still needs A's evidence. Removing A's adapter too early can make recovery impossible. A migration decision must account for that remaining work.
Name plausible consequences and the facts that would confirm them. Avoid unbounded speculation about every imaginable future. The scope and consequence of the actual change determine how far the analysis needs to go.
Inversion
Use this to challenge a proposed guarantee: describe a concrete way the promise could be violated, then look for the path that admits it. Untools: inversion.
For “only the reviewed version may publish,” try substituting a new version after approval, constructing an approval value directly, or replaying a stale decision after its authority expires. Each attempt targets a different relationship.
Some counterexamples should be excluded by type construction; others require current stored facts or source evidence. Inversion does not imply that every thought experiment becomes a permanent compiler-negative test. The project chooses its evidence.
Why These Tools
These six address framing, shared meaning, investigation, inference, change and counterexamples. Other tools may fit another question. A decision matrix is useful when alternatives and criteria are genuinely known; invented weights would only hide judgment here. A prioritization matrix does not decide a story's authority. No tool in the catalogue becomes a mandatory stage in every engineering task.