Evidence reasoning

Deterministic verification where possible. Probabilistic judgment where reality requires it.

If your decision depends on both hard facts and uncertain estimates, the review should keep those two types of evidence clearly separated.

Discuss a Matter View the Methodology

Reasoning pathways

Fixed Results and Uncertain Judgement

Possible route or outcome

01Known InputsVerified inputs and a fixed rule can produce a reproducible result.

02Fixed ResultThe output should be independently recalculable.

03Uncertain InputsIncomplete evidence requires probability, range or scenario analysis.

04Sensitivity and AlternativesTest how assumptions and competing explanations affect the result.

05Qualified ConclusionState confidence and reliance without presenting probability as certainty.

Two different tasks

Verification is not the same as judgment

Deterministic verification gates

Used where a matter can be tested against a defined record, condition, calculation or rule.

Identity, dates and chronology
Document and source existence
Citation accuracy and quotation context
Calculations and structural requirements
Directorships and necessary conditions
Whether apparent sources are genuinely independent

Structured probabilistic judgment

Used where several facts must be weighed and the record does not permit a binary answer.

Credibility of a representation
Strength of a market or transaction assumption
Significance of inconsistencies
Competing explanations
Materiality of evidence gaps
What further evidence could change the assessment

Illustrative situation

A source is correct but the conclusion is too broad

A draft report cites the correct document. Its summary extends the finding beyond the document’s stated conditions.

Does the final wording stay within the source’s actual support?

What we would examine: Check the original passage, limitations, dates and contrary material. Revise the finding and retain the relevant uncertainty.
What you could receive: A reviewed finding with traceable sources and an explicit reliance boundary.

How a finding could read, if the evidence supported it

The narrower proposition is supported. The broader conclusion has been removed because the reviewed evidence does not establish it.

Why this helps: You receive a conclusion whose scope can be inspected rather than a stronger sentence without stronger evidence.

This is an illustration, not a completed client assignment or a finding about a real organisation. The actual evidence determines the result.

Three separate questions

Factual status, evidential confidence and reliance status

1. Factual status

What can be directly verified, contradicted, located, calculated or left unresolved?

2. Evidential confidence

How strong, independent, complete and current is the support for the broader conclusion?

3. Reliance status

Is the material sufficiently strong for the intended decision, its consequences and the cost of being wrong?

The same evidence may justify different levels of reliance depending on the decision, the potential consequences and the cost of being wrong.

Confidence language

Qualified, explained and never invented

Aperture does not assign precise percentages unless a defensible statistical model exists. Broader assessments may use qualified confidence levels:

High confidenceStrong, independent and materially complete support, with no unresolved issue likely to change the central finding.

Moderate confidenceMeaningful support exists, but one or more material uncertainties or plausible alternatives remain.

Limited confidenceThe finding is tentative because support is incomplete, dependent, conflicting or materially constrained.

Insufficient basis for confidenceThe available record does not responsibly support a broader conclusion.

What Aperture does not imply

Deterministic verification does not mean that a person, company, transaction or future event can be proved safe, unsafe or certain in every respect. Probabilistic judgment does not mean speculation, intuition or an unsupported AI confidence score.

Probability is not presented as certainty, and uncertainty is not concealed by polished language or numerical confidence scores.

Reality before reliance

Know what can be verified, what remains uncertain and what the decision can responsibly bear.

Discuss a Matter Request a Formal Proposal