Section 301 of 440
Complete canonical tutorial. This reader section contains the same teaching body as PWR-119 · Decision under uncertainty. Open the Power dossier.
PWR-119 · SELF GUIDED full tutorial
Compare options on a fictional case using ranges, values and a decision threshold
A decision under uncertainty connects options with possible states, probability ranges, consequences, values, rights, reversibility and accountability. This lesson uses a fictional low-stakes venue choice. The learner predeclares objectives and non-negotiable constraints, estimates ranges, calculates one expected-cost comparison, checks sensitivity and records what new fact would change the choice. The worksheet must not choose a real clinical, legal, political, employment, financial or safety option.
1 · Permission and limits
Know exactly what you may do
2 · Get ready
Gather what you need and check the starting conditions
What you need
- Exact fictional venue card: accessible indoor room costs £180; courtyard costs £0 if dry and £500 if rain cancels; rain range is 30–50%; both options must retain the supplied step-free route. Two update cards say 25% and 45%, both with no access change; a helper chooses one face down and reveals it only after the switching threshold is committed.
- Option–state matrix, calculator and probability-range worksheet.
- Decision log with owner, affected roles, reversal and update trigger.
Before you start
- Write FICTIONAL PRACTICE at the top.
- Name the decision owner and protected constraints supplied by the case.
- Keep probabilities as ranges when evidence does not justify a point estimate.
3 · The method
Follow these steps in order
- Frame the decision
State the question, decision time, owner, affected roles and what success means.
Why: A clear frame prevents silent changes in objective or authority.
Check: One sentence names owner, options and deadline.
- List options and protected constraints
Write every available option and mark any accessibility, consent or safety requirement that no score may override.
Why: Some values are constraints, not prices in an average.
Check: Each option is checked against non-negotiable conditions first.
- Define uncertain states
List mutually distinct future states and probability ranges using the supplied base rate and evidence.
Why: Explicit states prevent vague best-case storytelling.
Check: Ranges are bounded and their assumptions are cited.
- Map consequences
For each option–state pair, record cost, delay, burden and who experiences it. Keep unlike outcomes in separate columns.
Why: A single total can hide distribution and value conflict.
Check: Every cell has concrete consequences and affected role.
- Classify reversibility and information value
Mark which fictional options can be trialled, delayed or reversed and what useful information each would produce before commitment. Record any cost of waiting.
Why: A reversible step can preserve options, but delay can also impose a real consequence.
Check: Each option has a reversal route, information gain and waiting cost or is explicitly marked irreversible.
- Calculate and test sensitivity
For each option, calculate expected cost by multiplying each state probability by that option’s cost in the state and adding the products: Σ probability(state) × cost(option,state). Use probabilities as decimals that sum to 1. Repeat at the low and high ends of the supplied range; do not average protected constraints.
Why: Sensitivity shows whether the choice depends on a fragile estimate.
Check: The record states whether the preferred option changes across the range.
- Choose a threshold and update rule
Write the current fictional choice, the reason, what probability or fact would switch it and who would review it.
Why: A predeclared threshold supports proportionate updating rather than hindsight.
Check: The final line includes choice, threshold, trigger and reviewer.
4 · Worked example
See the whole method used once
Scenario
A fictional club must choose an accessible indoor room for £180 or a free courtyard for an event; forecast rain is 30–50%, and cancellation costs £500.
Walkthrough
- Omar names the fictional coordinator as owner and wheelchair access as non-negotiable for both options.
- He lists indoor and courtyard and removes any courtyard setup that lacks the supplied accessible route.
- He defines rain and no-rain states with the 30–50% range.
- At 40%, expected cancellation cost for the courtyard is £200, above the £180 room cost; at 30% it is £150.
- He records that the choice switches at 180/500=36%; when the hidden card changes rain to 25%, he calculates £125 for courtyard cancellation, confirms the step-free constraint still holds and routes the possible switch to the fictional owner for review rather than acting automatically.
Result
Omar shows a threshold-sensitive fictional decision while preserving access as a constraint. The numerical result is not a recommendation for a real event.
5 · Right and wrong
Compare correct or safer execution with the common wrong version
| Moment | Right / safer | Wrong / riskier | Why it matters |
|---|---|---|---|
| Probability | Use a range tied to the supplied base rate. | Choose the most convenient point estimate. | False precision can determine the answer invisibly. |
| Values | Keep rights and access as constraints. | Assign them a small cost and average them away. | Not every legitimate value is tradable. |
| Sensitivity | Recalculate at both range ends. | Report one expected value as the optimal answer. | A fragile result may flip with plausible evidence. |
| Authority | Keep the result inside the fictional case. | Apply the worksheet to another person’s real treatment or job. | A calculation cannot grant decision authority. |
6 · Common mistakes
Spot the error and apply the correction
| Mistake | Fix |
|---|---|
| States overlap or omit an important possibility. | Make states mutually distinguishable and add a residual/unknown row. |
| Probabilities do not match the same time horizon. | Write the event and deadline beside every range. |
| All consequences are collapsed into money. | Keep burden, access, reversibility and affected roles in separate fields. |
| The update trigger is written after new evidence arrives. | Predeclare the switching threshold before revealing the update card. |
7 · Practice
Turn the steps into a usable skill
First session
- Frame one supplied fictional decision.
- Build the option–state matrix.
- Calculate a range-sensitive comparison.
- Apply protected constraints.
- Reveal one update card and follow the predeclared rule.
Repeat plan
Use one new fictional case weekly for four weeks, alternating cost, time and non-numeric consequences. On the fourth case, have another learner audit the authority, constraints and sensitivity rather than agree with the choice.
Progress when
- All options and states are explicit.
- The choice is stable or honestly labelled sensitive across the range.
- Updates follow a predeclared trigger.
Do not progress when
- A real consequential choice enters the worksheet.
- Affected people or protected constraints are missing.
- The learner treats one numerical model as the only legitimate value system.
8 · Check the result
Measure what changed
Completeness and internal consistency of a fictional uncertainty decision record.
How: Check owner, options, mutually exclusive states, probability range and source, consequences, affected roles, constraints, reversibility, threshold and update rule. Reproduce each expected-cost row as Σ probability(state) × cost(option,state) at both range ends and verify probabilities sum to 1.
Good result: All 12 fields are present, calculations reproduce, protected constraints are honoured and the update follows the declared trigger.
This does not prove: It does not identify an objectively optimal real choice, legitimate values, professional advice or safe high-stakes action.
Self-check
- Who owns the fictional decision?
- Which constraint cannot be averaged away?
- Does the choice flip within the probability range?
- What exact evidence triggers review?
9 · Stop, adapt or get help
Keep the safety boundary practical
Stop and get help
- Stop if the case becomes real, time-critical or consequential for another person.
- Stop if a model or AI supplies untraceable probabilities, hidden values or version-dependent advice.
- Route real high-stakes decisions to the legitimate accountable authority with appeal and affected-person participation.
Accessibility and adaptations
- Use a plain-language matrix with icons and spoken probability ranges.
- Allow calculator, spreadsheet or scribe support with assumptions visible.
- Separate one dimension per page to reduce cognitive load.
10 · Evidence and limits
Why these instructions are here
- primary research
Incorrect AI advice reduced human accuracy in a human-in-the-loop process, so advice source and version must remain visible.
The impact of AI errors in a human-in-the-loop process - official guidance
NIST AI RMF requires risk, context, governance and measurement around AI-supported decisions rather than assuming neutral optimisation.
Artificial Intelligence Risk Management Framework (AI RMF 1.0)
Limits
- Expected value addresses only declared quantifiable outcomes.
- Probability ranges and consequences can be wrong or contested.
- Fictional competence does not transfer automatically to real authority or high stakes.
Open the complete canonical research register
- Primary empirical supportLimiting / contraryThe effect of problem-solving and decision-making education on problem-solving and decision-making skills of nurse managers: A randomized controlled trial
Berra Yilmaz Kusakli; Betül Sönmez · 2024 · Primary research
- Primary empirical supportLimiting / contraryThe psychology of intelligence analysis: Drivers of prediction accuracy in world politics
Barbara Mellers; Eric Stone; Pavel Atanasov; Nick Rohrbaugh; S. Emlen Metz; Lyle Ungar; Michael M. Bishop; Michael Horowitz; Ed Merkle; Philip Tetlock · 2015 · Primary research
- Primary empirical supportLimiting / contraryTo Trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-assisted Decision-making
Zana Buçinca; Maja Barbara Malaya; Krzysztof Z. Gajos · 2021 · Primary research
- Primary empirical supportLimiting / contraryDebiasing Decisions
Carey K. Morewedge; Haewon Yoon; Irene Scopelliti; Carl W. Symborski; James H. Korris; Karim S. Kassam · 2015 · Primary research
- Primary empirical supportLimiting / contraryThe impact of AI errors in a human-in-the-loop process
Ujué Agudo; Karlos G. Liberal; Miren Arrese; Helena Matute · 2024 · Primary research
- Limiting / contraryOfficial boundary contextArtificial Intelligence Risk Management Framework (AI RMF 1.0)
National Institute of Standards and Technology · 2023 · Official framework
Read the complete evidence interpretation on the Power dossier.
Tutorial delivery controls
Learn, adapt, troubleshoot and resume
Progress is saved only in this browser on this device.
Step-by-step learner mode
Each activity includes its success check, a nearby accessible alternative and an “I’m stuck” correction path. Alternatives preserve the target where possible; when they change the task, Titan labels them as related rather than equivalent.
Frame the decision
State the question, decision time, owner, affected roles and what success means.
A clear frame prevents silent changes in objective or authority.
One sentence names owner, options and deadline.
I’m stuck on this step
Reset: Re-read this authored instruction — “State the question, decision time, owner, affected roles and what success means.” — and its success check, then attempt only this step.
Possible snag: The result from “State the question, decision time, owner, affected roles and what success means.” does not yet meet this declared check: One sentence names owner, options and deadline.
Correction: Return to the start of “Frame the decision”, reduce complexity or pace, and repeat only the part needed to satisfy: “One sentence names owner, options and deadline.”
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
List options and protected constraints
Write every available option and mark any accessibility, consent or safety requirement that no score may override.
Some values are constraints, not prices in an average.
Each option is checked against non-negotiable conditions first.
I’m stuck on this step
Reset: Re-read this authored instruction — “Write every available option and mark any accessibility, consent or safety requirement that no score may override.” — and its success check, then attempt only this step.
Possible snag: Probabilities do not match the same time horizon.
Correction: Write the event and deadline beside every range.
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
Define uncertain states
List mutually distinct future states and probability ranges using the supplied base rate and evidence.
Explicit states prevent vague best-case storytelling.
Ranges are bounded and their assumptions are cited.
I’m stuck on this step
Reset: Re-read this authored instruction — “List mutually distinct future states and probability ranges using the supplied base rate and evidence.” — and its success check, then attempt only this step.
Possible snag: States overlap or omit an important possibility.
Correction: Make states mutually distinguishable and add a residual/unknown row.
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
Map consequences
For each option–state pair, record cost, delay, burden and who experiences it. Keep unlike outcomes in separate columns.
A single total can hide distribution and value conflict.
Every cell has concrete consequences and affected role.
I’m stuck on this step
Reset: Re-read this authored instruction — “For each option–state pair, record cost, delay, burden and who experiences it. Keep unlike outcomes in separate columns.” — and its success check, then attempt only this step.
Possible snag: All consequences are collapsed into money.
Correction: Keep burden, access, reversibility and affected roles in separate fields.
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
Classify reversibility and information value
Mark which fictional options can be trialled, delayed or reversed and what useful information each would produce before commitment. Record any cost of waiting.
A reversible step can preserve options, but delay can also impose a real consequence.
Each option has a reversal route, information gain and waiting cost or is explicitly marked irreversible.
I’m stuck on this step
Reset: Re-read this authored instruction — “Mark which fictional options can be trialled, delayed or reversed and what useful information each would produce before commitment. Record any cost of waiting.” — and its success check, then attempt only this step.
Possible snag: The result from “Mark which fictional options can be trialled, delayed or reversed and what useful information each would produce before commitment. Record any cost of waiting.” does not yet meet this declared check: Each option has a reversal route, information gain and waiting cost or is explicitly marked irreversible.
Correction: Return to the start of “Classify reversibility and information value”, reduce complexity or pace, and repeat only the part needed to satisfy: “Each option has a reversal route, information gain and waiting cost or is explicitly marked irreversible.”
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
Calculate and test sensitivity
For each option, calculate expected cost by multiplying each state probability by that option’s cost in the state and adding the products: Σ probability(state) × cost(option,state). Use probabilities as decimals that sum to 1. Repeat at the low and high ends of the supplied range; do not average protected constraints.
Sensitivity shows whether the choice depends on a fragile estimate.
The record states whether the preferred option changes across the range.
I’m stuck on this step
Reset: Re-read this authored instruction — “For each option, calculate expected cost by multiplying each state probability by that option’s cost in the state and adding the products: Σ probability(state) × cost(option,state). Use probabilities as decimals that sum to 1. Repeat at the low and high ends of the supplied range; do not average protected constraints.” — and its success check, then attempt only this step.
Possible snag: The result from “For each option, calculate expected cost by multiplying each state probability by that option’s cost in the state and adding the products: Σ probability(state) × cost(option,state). Use probabilities as decimals that sum to 1. Repeat at the low and high ends of the supplied range; do not average protected constraints.” does not yet meet this declared check: The record states whether the preferred option changes across the range.
Correction: Return to the start of “Calculate and test sensitivity”, reduce complexity or pace, and repeat only the part needed to satisfy: “The record states whether the preferred option changes across the range.”
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
Choose a threshold and update rule
Write the current fictional choice, the reason, what probability or fact would switch it and who would review it.
A predeclared threshold supports proportionate updating rather than hindsight.
The final line includes choice, threshold, trigger and reviewer.
I’m stuck on this step
Reset: Re-read this authored instruction — “Write the current fictional choice, the reason, what probability or fact would switch it and who would review it.” — and its success check, then attempt only this step.
Possible snag: The update trigger is written after new evidence arrives.
Correction: Predeclare the switching threshold before revealing the update card.
Stop / get help: Stop if the case becomes real, time-critical or consequential for another person.
Correct versus incorrect execution
These accessible process diagrams are built from the tutorial’s own right/wrong teaching. They are not anatomical illustrations and do not add technique beyond the canonical tutorial.
Use a range tied to the supplied base rate.
Choose the most convenient point estimate.
Keep rights and access as constraints.
Assign them a small cost and average them away.
Recalculate at both range ends.
Report one expected value as the optimal answer.
Keep the result inside the fictional case.
Apply the worksheet to another person’s real treatment or job.
Method-structure checklist
10 of 10 structural checks present
- Ordered, Power-specific instructions — present
- Every activity has a success check — present
- Materials or supplied records are declared — present
- Measurement or assessment rule is present — present
- Tutorial-specific troubleshooting is present — present
- Stopping or escalation boundary is present — present
- Every activity has an adjacent alternative — present
- Correct-versus-incorrect comparison is present — present
- Evidence context is bound to the Power record — present
- Planning metadata is present — present
The method-readiness band and presence checklist assess tutorial presentation and are separate from evidence quality for the underlying Power. They are automated editorial aids, not human approval.
Manual editorial sign-off: Pending. This tutorial must not display a human-approved state until an identified editor signs the exact content hash.