Scenario Construction for Complex Systems: A Climate-Health Case Study
A couple of years ago I was challenged to think about methods for understanding the long-term implications of climate-health interactions. I was asked by a colleague to sort out some methods that would help public health planners understand the complexity of climate-health relationships and transform them into priorities for action. Data from current health outcomes (e.g. malaria, dengue, malnutrition, heatstroke) can be rare, especially among health ministries that aren’t functioning as knowledge networks. It is also common that methods supporting forecasting are viewed as impractical, confusing, and too complicated given that institutional systems are struggling to provide basic services – much less anticipation.
Because data about the status and direction of health outcomes can be notably absent, we focused our attention on scenarios and the different methodologies. Scenarios are valuable for health and technology, in part, because they contain a certain narrative closure. Clear winners and losers can emerge along with outcomes that measure conflict and contributions to the process. On the flipside, that narrative certainty is a little too clean. Real world interactions are messy. However, the most importune implication is that scenarios make good design tools because they suggest future arrangements and demonstrate alternatives without interfering in current practice. Scenarios shift the context to an indefinite time in the future, an aliased set of actors, or a new place to make new propositions less personal. This unbinds specific feelings of identity from new organizational arrangements and may leave participants free to experiment further.
Scenarios can be complicated to produce. They require focused study and time, and that seems too often in short supply. Plus, you need hooks to get people engaged in finding and discovering the elements that ought to belong. Scenarios should be plausible and internally consistent, but they also should be relevant to a broad range of stakeholders. Some methods focus too narrowly on their own visions of the world, and can end up decidedly deterministic or expertly biased, as this critique of Royal Dutch Shell’s approach explains (opens pdf).
Because the organization we were working with is committed to a open stakeholder process, we wanted a methodology that would accept diverse contributions and still be tied to one of the hallmarks of science: replicability. So we kept some design criteria in mind while we explored:
Scalable
We wanted techniques that could allow us to look scenarios for specific contexts and regions, from hospital units to watersheds and beyond.
Participatory
Being able to use many perspectives was a definitive goal. Not only are there differing accounts of actors and outcomes, participation does a much better job of revealing where goals might be in conflict in the system. Participation is also critical for helping the results of the scenario process diffuse among different stockholder groups.
Translatable across domains
Public health and complex systems are increasing supported by people and things from a variety of disciplines. We wanted insights from ethnographers to be as critical to the development of scenarios as live data streams of mechanical stress, if that’s what the scenario needed. We also wanted the materials and insights generated by the process to be amenable to visual display, since many of the stakeholders may use different languages. Visual formats also exploit the ambiguities of statements to reveal tensions that exist among interpretations.
Robust to diverse interpretations
Some of that tension is created when you get people from different backgrounds discussing what they think matters for interventions in particular health outcomes. Different levels of expertise can expose the assumptions that people share. The different elements of scenarios and how they emerge to affect long-term change often form the basis for many of this assumption. Highlighting this ambiguity is critical later for negotiating strategies for action.
Accepting of qualitative and quantitative insight
Working across disciplines is critical. One result of this is that the standards for evidence and data are different. We also recognize that quantitative measurement provides a detailed description of the identity or behavior of system elements. In particular, we wanted to be able to translate qualitative insights into format usable for compute modeling, simulation, and visualization.
Fun and pleasurable
Despite many people’s paradoxical notion that fun things are bad for you, we see fun as enhanced participation. When you forget that what you are doing is work, that’s a good thing.
Readily usable and modular
Methods should move seamlessly between health outcomes and altogether different domains. The process for malaria can be the same as heatstroke. Understanding alternative energy futures may use the same process as malnutrition. This enables practice and iteration.
As it turned out, scenarios techniques for climate-health interactions are not new, but they don’t deal well with uncertainty because they are explicitly aimed at extending interactions based on what the presence of domain knowledge and capable expertise. How could you hope to understand possible priorities and act all while not knowing? This was where we hoped to make a contribution.
Using Clamps to Build a Knowledge Network
Bob Johansen’s book, Get There Early outlines tools for dealing with dilemmas. Dilemmas confound rationality-based problem solving because of the way they are structured (multiple stakeholders, goals, conflicts, and outcomes, diverse framings and interpretations) and because there is not a clear path to one or a few positive solutions. Johansen outlines how Structure, Rules, Resources, Thresholds, Feedback, Memory, and Identity can be used as levers to help organizations attenuate themselves to the multi-textured shapes that dilemmas pose.
I think this list is pretty right-on for at least three reasons. First, the metaphor of levers directly brings to mind the work of Donella Meadows, an environmental scientist concerned with sustainability. Her work on leverage points for intervening in systems (pdf) is a great introduction and ordering of policy-based strategies and their efficacy for changing behavior. Like Johansen, she articulates the role of rules and feedback in systems. Meadows goes on to explore ten other significant systems levers, ultimately tracing effectiveness to how we frame the “problem”.
Second, structure, feedback, memory, and identity point to second order, emergent characteristics. Second order characteristics arise form the interactions of actors (e.g. people doing interesting things, wild coyotes, institutions, viruses), resources (e.g. coffee, water, land, low-interest loans, blood sugar), and their activities. Kevin Kelly explores why we are seeing more impossible events taking place. He connects it to an emergence of second order behaviors made possible through the development of new actors, new infrastructure, and new rules. Carl Simon, a Professor of Complex Systems at the University of Michigan, has studied the characteristics of complexity in biological and economic system and often differentiates complex behavior from simpler behavior by looking for heterogeneity, non-randomness, feedback, heterarchy, and emergence. Eric Berlow’s still great TED talk demonstrates how taking the broad, messy, and networked of complexity can in fact allow us to isolate clear paths for action.
The third reason I think Bob Johansen’s tuning levers are great is that they overlap with basic elements in game design. This should come as no surprise for most people associated with IFTF.
When I was working on the climate-health scenario methods, we faced a challenge of providing some sort of suitable structure for participants to embed meaningful insight into the scenarios. Sometime over morning coffee in a Swiss cafe, we stumbled across Tracy Fullerton’s rubric for the formal elements of games. These formal elements complement narrative elements and give rise to the more emergent properties of complex systems. Goals, procedures, actors, rules, resources, boundaries, conflicts, and outcomes also have a great synergy; they are exactly the elements used by computer programmers to construct agent-based models of complex adaptive systems!
Creating Relevance for Participation
So now we had a structural backbone for the kind of content we felt we needed to gather during a scenario development process. We could ask participants to engage in brainstorming activities that accounted for the different elements of these climate-health systems, and we would provide them with support, examples, and heuristics for doing just that. We also wanted to find a way to make the process fluid. In the back of our minds we always wanted to bring elements of game mechanics into the project to help support decision fatigue.
I’m still not sure we’ve cracked it, mostly because we haven’t been able to implement the process yet. However, we have looked at different forms of turn-based play with clear, articulated goals for the players, not unlike the LEARN, ACT, IMAGINE rubric that worked so well for Urgent Evoke missions.
One of the challenges is that we are introducing concepts about systems dynamics at the same time as concepts about the elements of the systems. This sets up a lot of material to get through in a short amount of time.
We also want to introduce experiences of empathy for others into the play and practice of scenario building. In order to generate robust scenarios, the goals of different actors represented need to be recognized and incorporated as valid contributions. One of the common experiences of public health service delivery is that managers, practitioners, patients, and others all have different views of the system. These occluded perspectives mean that they have a difficult time in finding ways to enhance the social and ecological resilience of infrastructure. I think if we had our choice, we would use experiences of empathy to reinforce principles along the lines of those championed by Nobel Prize winner Elinor Ostrom for designing long-enduring institutions.
Another significant outcome of clamps and elements for scenario development is that they clearly lend themselves to visual means of communication. Boundaries, resources, timings, and rules are common opportunities for change. Precise and ambiguous definitions can take on increased relevance, especially when dealt with creatively. One of the functions of mapping and visualization is to demonstrate this inherent ambiguity, pointing to areas for finding common ground. When we try to represent them visually, we are forced to make choices about the precise meaning of those boundaries, and this can be a significant source of cognitive dissonance for participants. But it’s exactly the form of dialogue that’s needed. It sets the stage for tactical strategies when conflicts emerge. Boundaries flow, and their meanings and borders can sometimes be adjusted to reach consensus or compromise.
Assembling Scenarios in Everyday Life
One of the questions designers (of scenarios, tools, artifacts, anything really) have to ask themselves is, “Where does this fit in everyday life?” One of the most useful rubrics I’ve come across for design is Products and Practices: Selected Concepts from Science and Technology Studies and from Social Theories of Consumption and Practice. (sorry, paywall). The authors make a case for a social and infrastructure-based approach to design. They identify acquisition (how we find it), scripting (how it shapes practice), appropriation (using it for something else), assembly (where we use it), normalization (sharing along with others), and finally practice (what activities it supports). What is great about this list is that it helps designers imagine the contexts of use.
In our scenario construction process we had to identify where this process existed along with a range of other activities that needed to be carried out by participants. This assembly meant that our process had to connect to other activities in a meaningful way.
The current processes and guidelines for conducting Vulnerability and Adaptation assessments in vulnerable regions hinge on their level of stakeholder involvement. Some processes are top-down, others bottom-up, and others a mix of expertise and engagement.
One way to assemble scenarios into these processes is to:
1) Define the scope and focus which usually means identifying the health outcome of interest.
2) Work out a baseline for which information may not exist. This is where defining system elements can be helpful for laying out current distributions and burdens, strategies for coping, early prioritization of “drivers”, and the interactions between elements that affect their dynamics.
3) From this point on, forecasts can be made about future trends and conditions. For example, what happens if boundaries change? How about if an actor appears or disappears?
4) Once forecasts are made, the task is to frame and narrate the interactions as scenarios. This is a great opportunity to develop the scenario through the eyes of others. Games, agent-based models, visualizations, and mapping can demonstrate change over time and the differences in scales affected while uncovering an array of interesting and unexpected interactions.
5) Isolation and sequencing asks participants to step back from what they produced, to look at the areas of concern, and to select the most relevant links between scenario elements. By focusing attention on these links, the next task is to order the steps they will need to affect change by listing priorities for action.
6) Package and disseminate the scenarios and the priorities for broad communication and feedback.
7) Use the feedback and resulting statements to assess how the scenario process and how it enabled participants to identify and act on the priorities they generated.
As you can see, it’s a richly-textured process, highly-amenable for visual communication, and ripe for engagement. I think one of the most important functions is the ability to expand the number of elements that matter to long-term change. One of the key decisions that participants have to make is to ask whether a resource, boundary, conflict, actor, rule, or procedure matters or makes a difference to the health outcome of interest. Here Gregory Bateson’s statement about information as, “a difference that makes a difference” looms large. More on that in the future.
Signals from Noise
One of the key endeavors of public health, infrastructure, and technology is the attempt to identify signals in noisy environments. Signals are utilized in biology to communicate across chemical gradients, metabolic networks, neuronal synapses, visual spectra, haptic musculature, individual displays of affection, and as invitation for cooperation across groups and societies. Technological systems stimulate behavior in new and exciting ways, but they can also script and normalize actions that may limit our abilities to find success.
The biggest challenges in generating signals for any medium is to make them relevant enough to transcend noise and competition from similar signals elsewhere. Synergistic timing with the individuals or groups receiving them is critical – as this will help them become meaningful in helping receivers revise their previous beliefs or come to new conclusions.
John Snow’s well-know map showing cholera cases in the London epidemic of 1854 clustering around the Broad Street well was an early success in distinguishing signals from noise using visualization and tight clamps that link actors (cholera, people, wells), boundaries (streets, houses), resources (water), and procedures (washing, drinking). These interactions clearly led to an understanding of a health outcome, and the relationships, once linked, could be used to forecast future scenarios.
PETLAB and the Red Cross/Red Crescent Climate Center have been collaborating to help illuminate different, contradictory signals, that may become confusing to recipients during a weather-based crisis. This game supports better decision-making to manage the damage of incorrect flooding predictions.
Before the Storm is another game from the Parsons/Climate Centre collaboration that introduces forecasting to new audiences and uses the scenarios produced to help identify what the participants feel would be the most relevant and practical stapes to take during a flooding emergency.
Climate Health Impact – a simulation based game designed to give biology students a better understanding of the health impacts of climate change. It does do a great job of representing standard practices worldwide that contribute to the understanding and management of emerging vectors. What I like here is the attention to new actors and their relationships with policy measures, research processes, and geography. There’s a lot of detail about disease specifics as well, but narratively, it does reinforce a fairly top-down perspective.
Agent-based models sometimes very effective for examining conflict among different actors. This paper by [img_assist|nid=3955|title=Hailegiorgis et al. models a human-environment interaction|desc=|link=none|align=right|width=199|height=104]Hailegiorgis et al. models a human-environment interaction (pdf) and demonstrates how cyclical rainfall can reveal a pattern of punctuated conflict. The pattern suggests that durable mechanisms for cooperation (e.g. clear boundaries, enforceable rules, mechanisms for redress, nested institutions) will be needed to traverse environmental change if the communities are going to maintain their resilience.
The Future of Scenarios
What do scenarios look like when the are disseminated and opened up for engagement? I think they look closer to everyday life. To understand the impacts of alternative scenarios we have to look at out interpersonal relationships – at the things that are one or two degrees removed. How will climate-health interaction affect our pets, our sex lives, how we eat dinner, getting to and from work, and our expectations when we encounter each other on the street? I think the genre of climate-health scenarios and perhaps all scenarios is not one of horror, western drama, or even fantastical sci-fi; it has to be more subtle, more internally embedded in social values and individual goals. It’s melodrama about how we live and how we live it everyday. That’s the real scary, far-out stuff.
. Or maybe the blogs are just difficult to find, in a different language, or just not cool enough. 
