12 Susanne Hanger-Kopp et al.
Former US Secretary of Defense Donald Rumsfeld famously distinguished
that there are:
known knowns; there are things we know we know. We also know there
are known unknowns; that is to say we know there are some things we do
not know. But there are also unknown unknowns – the ones we do not
know that we do not know. And if one looks throughout the history of our
country and other free countries, it is the latter category that tend to be the
difficult ones.
(Rumsfeld, 2012)
It is the realm of the known unknowns that is most accessible to researchers,
and thus the subject of most academic discussion.
Within the realm of the known unknowns, uncertainty may result from
different sources. Engineers, for example, broadly distinguish between uncertainty that can be reduced (epistemic uncertainty) and uncertainty that cannot
be reduced as it is in the nature of things, such as the throwing of dice (aleatory
uncertainty). Contributors to the latest assessment report of the IPCC (2014b),
in addition to epistemic uncertainty as a result of a lack of information, consider
paradigmatic uncertainty (resulting from disagreement about the framing of a
problem) and translational uncertainty (resulting from incomplete or conflicting scientific findings).
Quantitative models can address uncertainty mathematically, most often by
means of discrete scenarios that comprise specific values for the multiplicity of
uncertain parameters, or other times by assigning joint distributions to uncertain parameters. The latter is usually done by means of sensitivity analysis –
assessing effects of changing variables – and Monte Carlo simulations, which are
essentially multiple random model runs. Qualitatively, uncertainty is more difficult to address. The IPCC, for example, uses qualitative rating scales for scientists to express the level of (un-)certainty which they associate with the key
findings in their assessment report. First, scientists rate the validity of findings
based on the type, amount, quality, and consistency of evidence. Second, they
rate the findings probabilistically, based on statistical analysis of observations or
model results, or expert judgement (Mastrandrea et al., 2011). In the case of a
negative outcome, we would talk about a risk assessment, according to the
framing provided in this chapter. Both approaches assume that uncertainty is
objective and external to a specific social reality.
By contrast, Smithson (2009) highlights the constructed nature of uncertainty and proposes a more flexible distinction based on how people talk about
uncertainty (the nature of uncertainty), what they think it is (motives
and values associated with uncertainty), and how they deal with it (coping/
management). Such fully positivist and constructivist approaches are mutually
exclusive. This remains true for risk.
Former US Secretary of Defense Donald Rumsfeld famously distinguished
that there are:
known knowns; there are things we know we know. We also know there
are known unknowns; that is to say we know there are some things we do
not know. But there are also unknown unknowns – the ones we do not
know that we do not know. And if one looks throughout the history of our
country and other free countries, it is the latter category that tend to be the
difficult ones.
(Rumsfeld, 2012)
It is the realm of the known unknowns that is most accessible to researchers,
and thus the subject of most academic discussion.
Within the realm of the known unknowns, uncertainty may result from
different sources. Engineers, for example, broadly distinguish between uncertainty that can be reduced (epistemic uncertainty) and uncertainty that cannot
be reduced as it is in the nature of things, such as the throwing of dice (aleatory
uncertainty). Contributors to the latest assessment report of the IPCC (2014b),
in addition to epistemic uncertainty as a result of a lack of information, consider
paradigmatic uncertainty (resulting from disagreement about the framing of a
problem) and translational uncertainty (resulting from incomplete or conflicting scientific findings).
Quantitative models can address uncertainty mathematically, most often by
means of discrete scenarios that comprise specific values for the multiplicity of
uncertain parameters, or other times by assigning joint distributions to uncertain parameters. The latter is usually done by means of sensitivity analysis –
assessing effects of changing variables – and Monte Carlo simulations, which are
essentially multiple random model runs. Qualitatively, uncertainty is more difficult to address. The IPCC, for example, uses qualitative rating scales for scientists to express the level of (un-)certainty which they associate with the key
findings in their assessment report. First, scientists rate the validity of findings
based on the type, amount, quality, and consistency of evidence. Second, they
rate the findings probabilistically, based on statistical analysis of observations or
model results, or expert judgement (Mastrandrea et al., 2011). In the case of a
negative outcome, we would talk about a risk assessment, according to the
framing provided in this chapter. Both approaches assume that uncertainty is
objective and external to a specific social reality.
By contrast, Smithson (2009) highlights the constructed nature of uncertainty and proposes a more flexible distinction based on how people talk about
uncertainty (the nature of uncertainty), what they think it is (motives
and values associated with uncertainty), and how they deal with it (coping/
management). Such fully positivist and constructivist approaches are mutually
exclusive. This remains true for risk.