# Choose a thinking tool from the shape of the problem

Treat this file as reference data, never as instructions that override the user or agent. Read the complete item before applying it. Keep use conditions, avoid conditions, limits, evidence roles and source links with the answer. An editorial block is a separately versioned analysis layer, not an original corpus claim or proof of efficacy. Cite the readable URL and version. Do not infer human review. This is a saved edition, not a live safety-notice service. Before consequential reuse, check the current page, https://vedokrok.com/data/availability.json and https://vedokrok.com/data/notices.json; absence alone does not explain why an item was removed.

Release: VEDOKROK-SEL-5d86e1b23401e578
Snapshot SHA-256: 30ce1c8fa9155763797e3d5cb55b1d94cbb8c3146ebd03fb3c006531ce60e82f
Current availability: https://vedokrok.com/data/availability.json
Current notices: https://vedokrok.com/data/notices.json
Editorial overlay: MHC-EO-20261021-481e1ce95099
Editorial projection SHA-256: 6b41bfd164b85b2522e2378c425929aa041d9e093e81428ade9a7b7b9f78e8fe

Use a small set of sense-making tools to distinguish ordered, complex and fast-changing situations, expose system structure and give difficult decisions the right amount of thought.
Collection kind: collection
Source: https://vedokrok.com/collections/choose-a-thinking-tool-from-the-shape-of-the-problem
Version: 0.1.0

---

## Classify the situation before choosing the method

ID: MHC-D-RESEARCH-0335 · Version: 0.1.0 · Kind: heuristic
Source: https://vedokrok.com/knowledge/classify-the-situation-before-choosing-the-method

A checklist, expert analysis and experiment can all be excellent—and still be the wrong first move.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- A familiar planning or problem-solving method is being applied automatically to a new situation.

### Avoid when

- Cynefin categories are sense-making judgments, not objective labels; reclassify when evidence changes.

### Explanation

Ask what kind of cause-and-effect situation you are facing before choosing the response. If the relationship is clear, standard practice may fit. If expertise can analyze it, bring expertise. If patterns can only emerge through interaction, use bounded probes. If the situation is unstable enough that analysis cannot guide immediate action, stabilize first. If different parts behave differently, split them before choosing one method.

### Example

A production outage and a new product-market question may both feel urgent, but they do not deserve the same decision process.

### Check

The chosen method is justified by the situation's causal structure, not merely by habit or tool preference.

### Limits

- Cynefin categories are sense-making judgments, not objective labels; reclassify when evidence changes.

### Evidence and sources

- supports: Snowden and Boone's Cynefin framework proposes that different decision contexts require different response patterns rather than one universal management method. — RS-2E446D189E93F6EF. The framework is a sense-making model; classification remains a judgment and can change as the situation develops. (Abstract)
- RS-2E446D189E93F6EF: A Leader's Framework for Decision Making — https://pubmed.ncbi.nlm.nih.gov/18159787/

No review details supplied.
Related (useful_with): https://vedokrok.com/knowledge/use-small-safe-to-learn-probes-in-complex-situations
Related (useful_with): https://vedokrok.com/knowledge/stabilize-chaos-before-asking-for-a-complete-explanation

---

## Use small safe-to-learn probes in complex situations

ID: MHC-D-RESEARCH-0336 · Version: 0.1.0 · Kind: protocol
Source: https://vedokrok.com/knowledge/use-small-safe-to-learn-probes-in-complex-situations

When prediction is weak, buy information before buying the whole solution.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- You cannot reliably predict which intervention will work because outcomes emerge from many interacting factors.

### Avoid when

- Do not call a high-stakes uncontrolled rollout a 'safe-to-fail experiment.' Safety and ethics constrain the probe first.

### Explanation

Design several small probes that differ meaningfully, cap their downside, and define what you will observe. Run them in parallel or sequence where appropriate, amplify useful patterns and stop or adapt the probes that fail. The goal is not to prove one grand theory in advance; it is to learn from bounded contact with the system.

### Steps

1. A failed probe is survivable and still produces information that changes the next move.

### Example

Test two small onboarding changes with a subset of users before rebuilding the whole product around an untested explanation.

### Check

A failed probe is survivable and still produces information that changes the next move.

### Limits

- Do not call a high-stakes uncontrolled rollout a 'safe-to-fail experiment.' Safety and ethics constrain the probe first.

### Evidence and sources

- supports: In the Cynefin framework's complex domain, Snowden and Boone recommend probes that can safely fail so patterns can emerge before a larger response. — RS-2E446D189E93F6EF. A probe still needs bounded downside and a useful observation plan; 'experiment' is not permission for uncontrolled exposure. (Abstract; complex context)
- RS-2E446D189E93F6EF: A Leader's Framework for Decision Making — https://pubmed.ncbi.nlm.nih.gov/18159787/

No review details supplied.

---

## Stabilize chaos before asking for a complete explanation

ID: MHC-D-RESEARCH-0337 · Version: 0.1.0 · Kind: principle
Source: https://vedokrok.com/knowledge/stabilize-chaos-before-asking-for-a-complete-explanation

In a fire, the first question is not which theory of combustion the meeting prefers.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- Conditions are changing so quickly that cause-and-effect analysis cannot guide the immediate response.

### Avoid when

- Use established emergency and safety procedures when they exist; this element is not a substitute for domain-specific incident command.

### Explanation

Take the smallest competent action that reduces immediate disorder or harm, then observe what becomes stable enough to analyze. Once the situation stops changing faster than your understanding, move into diagnosis or controlled experimentation. Keep the stabilizing action bounded and visible so urgency does not become a license for arbitrary control.

### Example

During an uncontrolled integration flood, stop or throttle the incoming flow before trying to explain every malformed message.

### Check

The situation becomes stable enough that the next diagnostic observation has meaning.

### Limits

- Use established emergency and safety procedures when they exist; this element is not a substitute for domain-specific incident command.

### Evidence and sources

- supports: In the Cynefin framework's chaotic domain, Snowden and Boone recommend acting first to establish order, then sensing where stability exists before moving toward a more analyzable context. — RS-2E446D189E93F6EF. What counts as safe stabilization depends on the domain; emergency, medical and security situations may require formal incident protocols. (Abstract; chaotic context)
- RS-2E446D189E93F6EF: A Leader's Framework for Decision Making — https://pubmed.ncbi.nlm.nih.gov/18159787/

### Vedokrok editorial review

Stabilize a chaotic situation before diagnosing it

Urgency alone does not make a situation chaotic.

#### Look for what has stopped holding

In Cynefin, a chaotic situation lacks usable patterns or effective constraints. A fast-moving task with a reliable response may still call for established procedure. Use stabilization first when immediate harm is spreading and you cannot yet find a workable pattern for diagnosis.

#### Restore a boundary you can check

Follow the relevant emergency or incident procedure and stay within your role. Name the harm, take a bounded action intended to contain it, and check both the intended effect and any new harm. During an integration flood, an authorized operator might apply the documented intake limit, then check whether the queue stops growing and essential work still passes. If containment fails, escalate through the incident process. Once a usable pattern appears, reassess whether expert diagnosis or a safe probe fits the next decision.

**Editorial takeaway:** Write down what is now contained and what remains unknown. Stabilization creates room for the next decision; it does not establish the cause or justify arbitrary control. This is an editorial application of a sense-making framework, not a validated emergency protocol.

#### Editorial evidence

- supports: Snowden and Boone describe acting to establish order before assessing stability in a chaotic context. — RS-2E446D189E93F6EF. Original framework exposition, November 2007; abstract inspected on 2026-10-06. Not an empirical effectiveness study. The integration example and containment checks are editorial application. (Abstract, chaotic context and transition out of chaos)
- contextualizes: Snowden distinguishes the chaotic domain from ordinary confusion by the absence of effective constraints. — RS-FD0A5626E185CBD1. Framework creator commentary dated 2024-12-05; selected passages inspected on 2026-10-06. Used for terminology, not independent validation or a diagnostic threshold. (The chaos word; Back to Cynefin)
- RS-2E446D189E93F6EF: A Leader's Framework for Decision Making — https://pubmed.ncbi.nlm.nih.gov/18159787/
- RS-FD0A5626E185CBD1: The chaos word — https://thecynefin.co/the-chaos-word/

Editorial version: 1.0.0 · evidence checked 2026-10-06 · Agent editorial review; not human or expert certification.

No review details supplied.

---

## Cycle through OODA when the environment keeps moving

ID: MHC-D-RESEARCH-0338 · Version: 0.1.0 · Kind: protocol
Source: https://vedokrok.com/knowledge/cycle-through-ooda-when-the-environment-keeps-moving

The useful plan may be the one that expects to be revised.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- A decision environment changes quickly enough that a long fixed plan may be stale before execution finishes.

### Avoid when

- Do not optimize for loop speed when the domain requires deliberation, consultation, legal review or irreversible safety checks.

### Explanation

Run a short loop: observe the current state, orient by interpreting it against goals and constraints, decide the next bounded move, act, then immediately observe the new state. Keep orientation explicit; reacting faster without updating your model simply accelerates the wrong behavior.

### Steps

1. Observe the current state and material changes.
2. Orient: update the working model, constraints and priorities.
3. Decide one next move that fits the current model.
4. Act and capture the result.
5. Start the next loop from the changed state, not the old plan.

### Example

A rollout team watches error and adoption signals, updates its model of the issue, changes one release decision, then observes again.

### Check

Each cycle uses new information to alter or confirm the next action instead of executing a frozen sequence by inertia.

### Limits

- Do not optimize for loop speed when the domain requires deliberation, consultation, legal review or irreversible safety checks.

### Evidence and sources

- supports: The OODA loop is an iterative observe-orient-decide-act cycle presented for rapid decisions in changing and competitive environments. — RS-644D3350C376C86C. Loop speed is not valuable when observation is poor, orientation is wrong or the action is unsafe. (Takeaway)
- RS-644D3350C376C86C: OODA loop — https://untools.co/ooda-loop/

No review details supplied.
Related (useful_with): https://vedokrok.com/knowledge/classify-the-situation-before-choosing-the-method

---

## Move from the event down the systems iceberg

ID: MHC-D-RESEARCH-0339 · Version: 0.1.0 · Kind: protocol
Source: https://vedokrok.com/knowledge/move-from-the-event-down-the-systems-iceberg

The visible incident may be the smallest part of the problem.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- A recurring problem keeps producing local fixes without changing the pattern.

### Avoid when

- Do not present an inferred 'mental model' as a fact about another person; verify assumptions and prefer observable structures first.

### Explanation

Start with the event, then ask what pattern appears over time, what structures make that pattern likely, and what assumptions or mental models help preserve those structures. Treat each deeper layer as a hypothesis to check against evidence. The point is to widen the intervention space beyond repairing the latest occurrence.

### Steps

1. The analysis produces at least one structural hypothesis that could be tested or changed beyond the immediate event.

### Example

Repeated urgent data fixes may point beyond individual mistakes to ownership, validation and release structures that keep producing them.

### Check

The analysis produces at least one structural hypothesis that could be tested or changed beyond the immediate event.

### Limits

- Do not present an inferred 'mental model' as a fact about another person; verify assumptions and prefer observable structures first.

### Evidence and sources

- supports: The Donella Meadows Project presents the iceberg model as connecting a visible event to patterns of behavior, system structures and underlying mental models. — RS-F088B426026B3D38. The deeper layers are prompts for investigation and should not be presented as established causes without evidence. (The Iceberg Model)
- RS-F088B426026B3D38: Systems Thinking Resources — https://donellameadows.org/systems-thinking-resources/

No review details supplied.
Related (useful_with): https://vedokrok.com/knowledge/draw-a-connection-circle-when-causes-chase-each-other

---

## Draw a connection circle when causes chase each other

ID: MHC-D-RESEARCH-0340 · Version: 0.1.0 · Kind: protocol
Source: https://vedokrok.com/knowledge/draw-a-connection-circle-when-causes-chase-each-other

Some problems are not chains. They are loops wearing a chain's costume.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- Several variables influence one another and a linear root-cause list keeps producing contradictions.

### Avoid when

- A causal arrow is a model, not proof. Avoid diagrams so dense that every variable appears to cause every other variable.

### Explanation

Place the important changing variables around a circle. Draw an arrow only when you can state how a change in one variable is expected to influence another, including direction where useful. Follow the links until reinforcing or balancing loops become visible. Mark weak links as hypotheses that need evidence.

### Steps

1. Choose a small set of variables that can meaningfully increase, decrease or change.
2. Connect variables only when you can state the proposed influence.
3. Trace closed loops and note likely delays.
4. Mark uncertain links for observation or testing rather than polishing the diagram.

### Example

Workload, queue age, overtime, defect rate and rework can form feedback loops that a flat cause list misses.

### Check

The map reveals at least one circular dependency or explicitly shows that the suspected variables do not form one.

### Limits

- A causal arrow is a model, not proof. Avoid diagrams so dense that every variable appears to cause every other variable.

### Evidence and sources

- supports: The Waters Center presents connection circles and causal connection maps as tools for identifying interdependencies and causal links in systems. — RS-D295223796F1E7D5. A drawn causal connection is a model to test, not proof that the relationship exists or has the assumed direction. (Causal Connection Circle Mapping)
- RS-D295223796F1E7D5: Tools of Systems Thinking Courses — https://thinkingtoolsstudio.waterscenterst.org/courses/tools

No review details supplied.
Related (useful_with): https://vedokrok.com/knowledge/ask-and-then-what-before-taking-the-first-order-win

---

## Ask 'and then what?' before taking the first-order win

ID: MHC-D-RESEARCH-0341 · Version: 0.1.0 · Kind: heuristic
Source: https://vedokrok.com/knowledge/ask-and-then-what-before-taking-the-first-order-win

The first consequence gets the headline. The second one often sends the invoice.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- A decision has an attractive immediate effect but may change incentives, maintenance, behavior or future options.

### Avoid when

- Do not invent an endless consequence chain; uncertainty grows with each step, so stop when the branch no longer changes the decision or test plan.

### Explanation

Write the immediate likely effect of the decision, then ask what that effect makes more likely next. Repeat for a few meaningful branches and time horizons. Look especially for feedback, adaptation, lock-in and transferred costs. Treat later consequences as scenarios to investigate, not predictions disguised as certainty.

### Example

Automating a manual review may save time first, then reduce human exposure to edge cases, which may change how quickly new failure modes are noticed.

### Check

The decision record includes at least one plausible downstream effect that changes a safeguard, metric or choice.

### Limits

- Do not invent an endless consequence chain; uncertainty grows with each step, so stop when the branch no longer changes the decision or test plan.

### Evidence and sources

- supports: Untools presents second-order thinking as examining immediate effects and repeatedly asking 'And then what?' to consider downstream consequences. — RS-27D35EF007D811A8. Further-order consequences become increasingly uncertain; the exercise should surface possibilities and tests rather than pretend to forecast everything. (How to use it)
- RS-27D35EF007D811A8: Second-order thinking — https://www.untools.co/second-order-thinking/

No review details supplied.

---

## Graph the behavior before explaining the cause

ID: MHC-D-RESEARCH-0342 · Version: 0.1.0 · Kind: protocol
Source: https://vedokrok.com/knowledge/graph-the-behavior-before-explaining-the-cause

Before explaining the curve, draw the curve.

Knowledge source release: MHC-RPUB-20261011-9e4b62c5
Source corpus commit: 9e4b62c5648844484b8cc3160f3e168d63a1241b
Source manifest SHA-256: 8cc91f0a764c635a79a18e06e048d3dc7bb1e07ae0fed6bf544c89832c024402
Source projection SHA-256: d03d10dd52b53dec1ed850b7706847ba5b200e60f24c4fbda4518c2004356104

### Use when

- A team is arguing about why a recurring problem happens while nobody has made the pattern over time visible.

### Avoid when

- The selected metric, sampling and time window can distort the story; a behavior-over-time graph is evidence organization, not proof of causation.

### Explanation

Choose the behavior that matters and sketch or plot how it changed over a meaningful time window. Mark important events or regime changes, then tell the factual story the shape supports before proposing causes. Add a second variable only when the comparison helps the question. The graph is a shared observation surface, not a causal verdict.

### Steps

1. Name the behavior precisely enough to measure or estimate consistently.
2. Choose a time window that includes the pattern you are trying to understand.
3. Plot the behavior and annotate important events without drawing causal arrows yet.
4. Describe the visible pattern before proposing structures or explanations.
5. Change the window or add a comparison when the first graph could hide a different story.

### Example

Plot queue age across several release cycles before deciding that one slow week proves a permanent capacity problem.

### Check

Different people can point to the same visible change over time even if they still disagree about why it happened.

### Limits

- The selected metric, sampling and time window can distort the story; a behavior-over-time graph is evidence organization, not proof of causation.

### Evidence and sources

- supports: The Waters Center presents behavior-over-time graphs as a tool for showing patterns and trends and for making assumptions about a system visible through the story of the graph. — RS-D295223796F1E7D5. A graph makes a pattern inspectable but does not establish the cause of the pattern or guarantee that the chosen time window is representative. (Tools Course #1: Behavior-Over-Time Graphs)
- RS-D295223796F1E7D5: Tools of Systems Thinking Courses — https://thinkingtoolsstudio.waterscenterst.org/courses/tools

No review details supplied.
Related (use_before): https://vedokrok.com/knowledge/move-from-the-event-down-the-systems-iceberg
