Commented abstract
The entry offers a compact but decisive definition: uncertainty is deep when actors do not agree on models, probabilities, or the evaluation of outcomes. It is not a mere lack of data nor a difficulty of calculation, but a condition in which the very structure of the question does not yet permit a stable probabilisation.
Structured commentary
Introduction
The merit of the entry is its definition: uncertainty is deep when the disagreement bears not on the values of parameters but on the generative models, the probabilities, or the criteria for evaluating outcomes. It follows that it does not reduce to a lack of data nor to a computational difficulty; it is a condition in which the structure of the question precludes, for the moment, a stable probabilisation. For MARTRO this clarification protects the method from a frequent drift — the conversion into number of what is not yet ready to be measured as risk.
The relevance for the small firm is immediate. When it evaluates a management system, an expansion, a key hire, or an automation, it seldom possesses processes stable enough to estimate their effects precisely: historical data reflect a configuration in flux, preferences among speed, control, quality, and cash are not always explicit, and internal actors may read the same problem in divergent ways. In such conditions, numerical precision risks functioning as premature reassurance.
The MARTRO translation is, first of all, classificatory: before choosing the instrument, one classifies the type of ignorance. If the problem is measurable risk, one resorts to data and calculation; if it is an alternative scenario, one reasons in terms of robustness; if it is deep uncertainty, one preserves options, reduces irreversibility, and defines revision rules. The editorial value of this classification is strong, for it teaches that not all problems merit the same language.
On the operational plane the entry lends itself to becoming a small grammar of decision, articulated in a few questions: which implicit models are we adopting? Who dissents? Which information is effectively lacking? Which decision becomes irreversible if we proceed now? What can we undertake on a reduced scale to learn without foreclosing options? Questions that withdraw uncertainty from rhetoric to place it within the architecture of the choice.
The boundary consists in preventing "deep uncertainty" from degrading into an alibi for inaction. The correct posture is not the suspension of action but the adjustment of commitment to the degree of knowledge: where data are adequate, a quantitative evaluation is obligatory; where they are not, the diagnosis must indicate which signals to render observable, which margins to preserve, and which stop rules to formulate.
Why it matters for MARTRO
it furnishes the canonical definition of deep uncertainty and separates it from mere lack of data, preventing premature probabilisation.
Limits and boundaries of use
it is a definitional entry; its value depends on the classification that follows.
deep uncertainty does not justify inaction but the adjustment of commitment to knowledge.
Practical application for SMEs
classify the type of ignorance before choosing the instrument (calculation, robustness, options).