construct of reality and, as such, are not absolute; thus, presenting a range
of plausible outcomes is not only desirable but justified. Nevertheless, there
is a greater burden on the scientist to build a better model, to increase confidence in the model, and to communicate the applicability and limitations
of the model in a more transparent manner.
5.2.14 Decision Context
What to do when being sued? The discussions of uncertainty, use of
scenario and sensitivity analyses, extrapolation, and other issues discussed
above are all affected by the nature of the decision process. One obvious
case is where the model is to be used to produce predictions for an adversarial or litigation process as opposed to being used by a decision maker to
choose among alternative options. Sometimes, the confidence burden is
much higher because an adversary can attempt to challenge the veracity of
the model. And models are ripe for attack by the very nature of the uncertainties, extrapolation issues, complexity, and use for predicting results
outside the range of the experimental evidence. For example, it would be
easy to argue that data based on one species may be wrong when applied
to another, even taxonomically similar, species for the simple reason that
cases can be found where that extrapolation gives wrong results. On the
other hand, in many cases, a particular model is widely used in litigation,
even when it can be demonstrated to be misleading or inappropriate,
because of the precedent of using that model.
5.2.15 Library of Case Studies
How to build on other experiences? One important mechanism to counteract some of the issues related to extrapolation, model inconsistencies, or
living with uncertainties is the development of a library, or at least a bookshelf, that documents the use of complex models in complex environmental decision-making situations. Ideally, when a sufficient number of such
case examples are documented, a decision maker will have clear guidance
on the issue at hand and on the utility of models to address that issue. For
example, a wide range of models and their results (e.g., hydrodynamic,
stream flow, soil erosion, ecotoxicology, and ecosystem) have been incorporated into regulatory policy during the past three decades. The library
should include not just these successes but also failures, where wrong
decisions were made or where models were found to be incorrect in their
predictions. Modelers tend not to publish failures, and, for that matter,
decision makers do not often publicize mistakes. Yet these outcomes are
often opportunities to reexamine critically the model construct and assumptions, resulting in a much improvement. The ultimate confidence in models
used in the decision processes will ensue when there is a sufficient record
of successes and clear guidance on what not to do.
5. Overcoming Barriers to the Use of Models
103
of plausible outcomes is not only desirable but justified. Nevertheless, there
is a greater burden on the scientist to build a better model, to increase confidence in the model, and to communicate the applicability and limitations
of the model in a more transparent manner.
5.2.14 Decision Context
What to do when being sued? The discussions of uncertainty, use of
scenario and sensitivity analyses, extrapolation, and other issues discussed
above are all affected by the nature of the decision process. One obvious
case is where the model is to be used to produce predictions for an adversarial or litigation process as opposed to being used by a decision maker to
choose among alternative options. Sometimes, the confidence burden is
much higher because an adversary can attempt to challenge the veracity of
the model. And models are ripe for attack by the very nature of the uncertainties, extrapolation issues, complexity, and use for predicting results
outside the range of the experimental evidence. For example, it would be
easy to argue that data based on one species may be wrong when applied
to another, even taxonomically similar, species for the simple reason that
cases can be found where that extrapolation gives wrong results. On the
other hand, in many cases, a particular model is widely used in litigation,
even when it can be demonstrated to be misleading or inappropriate,
because of the precedent of using that model.
5.2.15 Library of Case Studies
How to build on other experiences? One important mechanism to counteract some of the issues related to extrapolation, model inconsistencies, or
living with uncertainties is the development of a library, or at least a bookshelf, that documents the use of complex models in complex environmental decision-making situations. Ideally, when a sufficient number of such
case examples are documented, a decision maker will have clear guidance
on the issue at hand and on the utility of models to address that issue. For
example, a wide range of models and their results (e.g., hydrodynamic,
stream flow, soil erosion, ecotoxicology, and ecosystem) have been incorporated into regulatory policy during the past three decades. The library
should include not just these successes but also failures, where wrong
decisions were made or where models were found to be incorrect in their
predictions. Modelers tend not to publish failures, and, for that matter,
decision makers do not often publicize mistakes. Yet these outcomes are
often opportunities to reexamine critically the model construct and assumptions, resulting in a much improvement. The ultimate confidence in models
used in the decision processes will ensue when there is a sufficient record
of successes and clear guidance on what not to do.
5. Overcoming Barriers to the Use of Models
103