Why this matters now
Important decisions rarely arrive with perfect information.
Contexts change.
Evidence is incomplete.
Time is limited.
Stakeholders disagree.
New information appears after action has already started.
Generative AI adds a further complication.
A response can sound complete before the evidence behind it is complete.
This creates two weak strategies.
Wait indefinitely for certainty.
Or act with more confidence than the evidence justifies.
Responsible judgement requires a third option:
move forward while remaining honest about what is uncertain.
Real-Life Readiness captures the principle directly: being prepared does not mean having certainty; it means being able to move responsibly without it.
Uncertainty does not mean knowing nothing
A team can know a great deal and still face uncertainty.
One variable may be missing.
One claim may remain unverified.
Two explanations may still be plausible.
A stakeholder perspective may be absent.
The future response of a system may be difficult to predict.
The first task is therefore not to “remove uncertainty”.
It is to make its structure visible.
What do we know?
What supports it?
What are we assuming?
What needs checking?
What do we know we do not know?
What might we still be missing entirely?
Four useful states
EVIDENCE
What is sufficiently supported to enter our judgement.
ASSUMPTION
What we are temporarily treating as true in order to reason or proceed.
UNVERIFIED
A plausible claim that still requires checking.
UNKNOWN
Something we recognise that we do not yet know.
These are working distinctions, not a universal scientific taxonomy.
Their purpose is to prevent plausibility from quietly becoming fact.
| State | Question |
|---|---|
| Evidence | What supports this? |
| Assumption | What are we temporarily accepting? |
| Unverified | What still needs checking? |
| Unknown | What do we know we do not know? |
Confidence should reflect support
Good judgement does not need every statement to sound certain.
It can distinguish:
strong confidence;
moderate confidence;
weak confidence;
open uncertainty.
This matters because decision-makers frequently need to communicate both:
what they currently believe
and
how strongly the evidence supports that belief.
Plausible certainty
Generative AI makes a particular danger easier to see.
A response may be:
fluent;
well organised;
detailed;
apparently authoritative.
None of those properties guarantees that the underlying evidence is complete.
Real-Life Readiness describes the risk of plausible certainty: people and technologies can generate answers that appear complete before the evidence is complete.
The relevant question changes from:
“Does this sound convincing?”
to:
“What supports it, and what could still be missing?”
How much certainty do we need?
This is the practical centre of the concept.
DECIDE is not defined as reaching certainty.
It is defined as determining what is reliable enough to act on.
The threshold should depend on the decision.
Trying a new meeting structure for a week is not equivalent to making an irreversible infrastructure choice.
Different commitments justify different levels of verification.
Reversibility changes the threshold
Real-Life Readiness explicitly includes distinguishing reversible and irreversible decisions as part of acting under uncertainty.
This supports a useful design principle:
the harder a decision is to reverse and the greater its consequences, the stronger the justification should be before commitment.
Reversible actions can often be used to learn.
Irreversible actions require more evidence before they become real.
| Consequence / commitment | Lower evidence confidence | Higher evidence confidence |
|---|---|---|
| Lower consequence · reversible action | Run a bounded test to learn. | Act proportionately and monitor the results. |
| Higher consequence · hard-to-reverse action | Verify further before committing. | Confirm context, consequences and safeguards before proceeding. |
Conceptual guidance: not a score or an automatic threshold for approval.
Action can create evidence
Sometimes the best next move is not:
decide A
or
decide B.
It is:
run a bounded test that helps us decide.
Prototype.
Pilot.
Interview.
Experiment.
Simulation.
Limited implementation.
This links decision-making under uncertainty directly to Experiential Learning.
Action becomes a way of learning, not simply executing.
Not every unknown needs to be resolved
Information has a cost.
Research takes time.
Attention is limited.
A useful question is:
Would knowing this change what we choose?
If not, that uncertainty may not deserve immediate attention.
The objective is not total knowledge.
It is decision-relevant knowledge.
Ask what would change your mind
A robust decision should be able to identify possible review triggers.
What evidence would make us stop?
What result would make another option stronger?
What assumption, if disproved, would undermine the decision?
If there is no conceivable evidence that would change our view, we may no longer be judging.
We may simply be defending.
- Decision
- Supporting evidence
- Critical assumptions
- New evidence that triggers review
Revision is not failure
New evidence can justify a new decision.
Real-Life Readiness explicitly includes revising as evidence changes.
This matters culturally.
Consistency should mean consistency of reasoning and responsibility.
Not permanent loyalty to the first answer.
Some uncertainty is about values, not facts
More information cannot resolve every difficult decision.
Sometimes we understand the facts and still have competing values.
Privacy versus security.
Speed versus inclusion.
Autonomy versus protection.
Present benefit versus future capacity.
DECIDE explicitly includes competing values and trade-offs.
Evidence informs these choices.
It does not automatically make them for us.
Trade-offs should be visible
A decision should be able to explain:
what it prioritises;
what it sacrifices;
which risk it accepts;
which alternative it rejects;
why.
A choice presented without trade-offs often hides part of the actual decision.
Missing perspectives are also uncertainty
A dataset can be large and still incomplete.
A team can have deep expertise and still possess a blind spot.
A missing discipline.
A missing user.
A missing community.
A missing form of lived experience.
Real-Life Readiness includes searching for missing perspectives as part of moving responsibly under uncertainty.
This connects the topic to Interdisciplinarity, Neurodiversity and Ancestral Knowledge.
Decisions are not predictions
Prediction asks:
What will happen?
Decision asks:
What should we do given what we know now?
Perfect prediction is not required for a defensible decision.
But the decision should understand where its uncertainty lies and how it will respond if reality behaves differently.
Good outcome does not necessarily mean good decision
A weak process can get lucky.
A strong process can encounter an unexpected outcome.
Decision quality should therefore include:
evidence used;
assumptions identified;
alternatives compared;
trade-offs examined;
consequences considered;
review conditions defined.
Results matter.
But they are not the only evidence of judgement.
Relationship to Agency
The concepts should remain separate.
Decision-Making Under Uncertainty explains how judgement is built when evidence remains incomplete.
Agency is broader.
It asks whether a person can turn understanding and judgement into choice, action, authorship and responsibility.
One prepares the decision.
The other owns it.
Relationship to CARE
Decision does not end responsibility.
Once action begins, reality generates evidence.
Consequences appear.
CARE requires people to remain accountable for what follows.
Observe.
Respond.
Protect.
Revise.
Uncertainty continues after implementation.
An operational framework
- Known
- Unknown
- Assumptions
- Verify
- Options
- Reversibility
- Trade-offs
- Decision
- Consequence
- Feedback
- Revision
KNOW
What is supported?
UNKNOWN
What do we still not know?
ASSUMPTIONS
What are we temporarily taking for granted?
VERIFY
What needs checking before commitment?
OPTIONS
What legitimate alternatives exist?
REVERSIBILITY
What can be undone?
TRADE-OFFS
What does each option gain and sacrifice?
THRESHOLD
What level of evidence is sufficient for this level of consequence?
DECIDE
Choose and own a direction.
OBSERVE
What happens?
REVISE
What new evidence changes the judgement?
The complete movement is:
EVIDENCE → JUDGEMENT → COMMITMENT → CONSEQUENCE → LEARNING
The Decision Threshold Test
Before acting:
What do we know?
How do we know it?
What are we assuming?
What remains unverified?
What missing information could change the decision?
Whose perspective is missing?
Is the decision reversible?
What is the cost of waiting?
What is the cost of being wrong?
What would trigger review?
These questions do not generate an automatic answer.
They make the judgement inspectable.
What decision-making under uncertainty is not
It is not guessing.
It is not evidence-free action.
It is not waiting for perfect certainty.
It is not blind trust in intuition, experts or AI.
It is not endless analysis.
It is not the elimination of risk.
And it is not using uncertainty as an excuse to avoid responsibility.
It is the practice of matching commitment to evidence and consequence.
What the evidence says
Real-Life Readiness describes real problems as situations where information may be incomplete, time and resources limited, stakeholders may want different things and new evidence may appear after a decision.
Its response is to identify what is known, name what is unknown, separate evidence from assumption, distinguish reversible from irreversible decisions, test claims, seek missing perspectives, decide and revise when evidence changes.
DECIDE integrates those conditions as judgement with incomplete information, competing values and trade-offs, determining what is reliable enough to act upon.
This supports the conceptual framework.
It is not a validated psychometric measure of decision-making under uncertainty.
What this means in Mars Challenge
Mars Challenge deliberately places learning inside open problems, constraints and consequences. Its 2026 framing describes the programme as an experiential environment for human decision-making under planetary limits.
Uncertainty becomes visible when teams:
frame an incomplete problem;
evaluate evidence;
use AI;
choose what to research;
decide what to prototype;
compare alternatives;
receive feedback;
revise.
The learning opportunity is not simply to produce the winning answer.
It is to make visible how an answer became a decision.
This describes programme architecture rather than measured causal outcomes.
What this means for educators
If learners are always given all relevant information, decision-making under uncertainty remains invisible.
Learning environments can introduce bounded uncertainty safely.
Incomplete evidence.
Time limits.
Claims requiring verification.
New evidence midway through the process.
Competing stakeholder needs.
Reversible and irreversible options.
Then ask:
What did you know?
What did you assume?
What did you verify?
What did you decide not to verify?
Why was the available evidence enough?
What would have changed your decision?
The objective is not to reward lucky prediction.
It is to make judgement visible.
Related concepts
Projects and evidence in practice
Mars Challenge — applied decision environment
Open problems, investigation, constraints, prototyping, feedback and revision.
Evidence status: documented programme architecture; not a causal outcome claim.
Frequently asked questions
Can we make a good decision without certainty?
Yes. Many real decisions require action before complete certainty is available. Quality depends on understanding the evidence, acknowledging unknowns, exposing assumptions and matching commitment to consequence.
How much evidence is enough?
There is no universal threshold. A reversible, low-consequence test may justify action earlier. An irreversible or high-consequence commitment should normally require stronger evidence and broader verification.
Is uncertainty the same as being unprepared?
No. Uncertainty belongs to reality. Being unprepared means being unable to recognise, calibrate or act responsibly within it.
Should we always wait for more information?
No. Waiting also has costs. Good judgement compares the cost of acting too early with the cost of delaying action.
What changes with AI?
AI can make uncertainty harder to see because plausible responses can appear complete. Human judgement must therefore examine sources, assumptions, missing context and verification needs before converting output into action.
Does changing a decision mean the original decision was wrong?
Not necessarily. The original choice may have been reasonable given the evidence available. Responsible judgement includes updating when circumstances or evidence change.
How is this different from Agency?
Decision-making under uncertainty focuses on building judgement. Agency covers the wider movement from judgement to choice, action, authorship and responsibility.
References
Real-Life Readiness — A Foundational Framework
Adelino Sousa / Mars Challenge / Virtual Educa, 2026.
Uncertainty, plausible certainty, reversibility, DECIDE and revision.
Mars Challenge 2026 — Earth
Mars Challenge, 2026.
Programme framing around human decision-making under constraints and planetary limits.
Explore next
Next step
Deepen: Explore Agency, Real-Life Readiness and Evidence.
Explore the project: Examine where investigation, decision, feedback and revision appear in Mars Challenge.
Act: Apply the Decision Threshold Test to a real decision.
Updated: 14 October 2026