Zellner, M.L. (2008). Embracing complexity and uncertainty: The potential of agent-based modeling for environmental planning and policy. Planning Theory and Practice, 9.4: 437-457.
Zellner represents the field of urban planning and policy. However, she is most interested in environmental policy and planning. To Zellner, environmental planning is a dynamic field that has shaped its analytical and policy approaches over the decades. For instance, social movements in the 1960s led to a promise of engineering solutions as a means to solve environmental problems (through regulatory and technological approaches in the 1970s). Comprehensive urban modeling fell into disrepute in the 1970s, so “systems theory fueled the development of mathematical models applied to natural resource management” (pg. 437). In the 1980s, there was a rise in economic analyses and valuation methods related to economic goods and services—in other words, cost-benefit analyses for policy formation. Also in the 1980s there were GIS technologies for environmental impact assessments and identification of ecologically sensitive areas. While environmental policy has shifted along with changes in technology and values, issues are still complex. Zellner claims that small changes are often what are needed to properly maintain and sustain resources. Complexity presents a number of challenges, but to Zellner, complexity can be viewed as an advantage. “Agent-based modeling has the potential to support this kind of thinking, though this potential has yet to be fully realized in environmental planning and policy” (pg. 438).
“Agent-based modeling, widely used in the study of complexity, natural resources, and increasingly social sciences, is an alternative method that explicitly represents complexity and treats uncertainty in our analyses. These models are typically object-oriented compyter programs, where the objects are actors operating at various scales (e.g., residents, farmers, businesses, units of government) making rule-based decisions in an environment characterized by various attributes (e.g., presence of natural amenities, distance to employment centers, access to water and sewerage infrastructure)” (pg. 439).
Agent-based models are intended to serve as representations of complexity. They are also meant to serve as metaphors in that real-life actions and natural processes are represented as analogous mechanisms in the models. ABMs are meant to serve as “comparative devices” that can be used to explain and understand phenomena, “particularly if they are intuitive to a broad audience that includes scientists across the disciplines, policy makers, and community members” (pg. 439). To Zellner, ABMs can help facilitate collective action and knowledge, leading to desired environmental outcomes. As such, ABMs serve as participatory planning support systems “because they organize complex information into meaningful and accessible forms and identify the key actions that can make a difference” (pg. 440). The role of modeling, according to Zellner, remains external to the policy process (see Figure 1). Further, model assumptions and alternative goals are rarely discussed prior to or in conjunction with the assessment of means (pg. 442).
In the traditional analytical approach, subjective discussions of value judgement are kept separate from objective modeling activity, yet a normative decision is taking place that is not being openly discussed (pg. 442). In a traditional model such as this one, “policy makers request scientists to predict best estimates to guide policy decisions in response to constituents’ demands” (pg. 442). To further this problem, scientific research and participatory planning have issues; scientific research is unaware of social and cultural context, while participatory planning is unaware of the informational substance to assist in discussions of desirable goals and means (pg. 443). The goal of most modeling is prediction, much like classical scientific research attempts to produce estimates based on observed pattern changes. Zellner notes that models based on scientific research can be unintelligible and alienating, especially for citizens who have a stake in decisions about the environment. In short, predictive models present a “black box” to most citizens, thus discouraging them from contesting or even engaging with said models.
To conclude her article, Zellner suggests how agent-based models, as they operate as more intelligible, metaphorical representations of risk, are a solution to better engage citizens and stakeholders. As a result, the authority of scientific knowledge (or at least the single source of that knowledge) is contested, and more knowledges are valued and offered through agent-based models. Overall, it can be said that agent-based models are a transformational and democratizing solution to traditional analytical approaches to policy. While Zellner does not describe or mention “citizens” per se, she does identify “stakeholders” and has a clear objective to ensure that policy formation represents knowledge of the many instead of the one. Despite this hopeful tone, however, Zellner’s method would still follow a one-way Jeffersonian model (see Waddell, 1995) where stakeholders have a right to participate, though they may or may not contribute to the policy making process (or intervene in any way). This is still an improvement on the status quo of most traditional policy models, however. To offer an improved model, Zellner suggests a revised, alternative approach to integrated analysis and policy making. This alternative approach seems more in line with Waddell’s social constructionist model of public deliberation.