Integrated Assessment and Integrated ModelsIntegrated modeling is a particularly powerful and general tool for conducting integrated assessment but is not identical to integrated assessment, though the two are often confused or the terms used interchangeably. Most assessment projects now under way are developing an integrated model, and many are putting most of their effort into the model, while most commentators on integrated assessment are careful to assert that a model is not sufficient. Although several different relationships between assessment and modeling are possible, in past projects there has been little correlation, positive or negative, between how strongly modeled a project was and its effectiveness.
The essence of integrated assessment is providing a systematic way of integrating knowledge across disciplines, styles, resolutions, and degrees of certainty. This requires a coherent means of defining and meeting information needs across these borders, in useful form and in time. Building an integrating model may be the most effective way of imposing this discipline of consistency and mutual intelligibility across subdomains of the problem. Other incentives may also favor using a model to do the integration. Because a model is an identifiable product, making one may be an easier activity to promote to funders than other kinds of assessment. Because the challenges of model building are more specific, more technical, more readily bounded and easier to talk about, more strongly modeled assessments may also be attractive to researchers.
But the beneficial discipline of a single integrating model also carries costs. The rigor of forcing knowledge from disparate domains into a single formal model can become Procrustean, involving falsifying sectoral information, imposing inappropriate restrictions, and yielding aggregate results that say as much about algorithmic artifacts as they do about the science of the components.
Other ways of integrating an assessment than through an integrating model are also possible, though at the cost of imposing less stringent standards of cross-disciplinary consistency and intelligibility. Assessments can be developed by judgmental integration of expert knowledge across the relevant fields even with no formal modeling, as has been done in many studies of the Office of Technology Assessment (OTA). Alternatively, assessments can use formal models of sub-components of the problem, linked through judgmental combination of their results rather than through a formal integrating model, as in the MINK (Missouri-Iowa-Nebraska-Kansas region) and CIAP (Climatic Impact Assessment Program) studies.
Even if an integrated assessment uses a formal integrating model, the model can play many roles in the overall assessment. Building the model may be the central task of the project and command essentially all resources. Alternatively, areas of the problem may be studied on parallel tracks, with one track developing a representation for the model and others examining the area in greater depth. Even a project centered on a model may decide to study some areas that are unlikely to be representable in the model, simply because they are deemed sufficiently important.
The next page is Flexibility, Abstraction, and Accessibility.