This presentation provides an overview of the Elephant Builder tool, developed by Bellwether Collabatory. This was recently showcased at the Resilience Shift's tools and approaches workshop.
Whether you’re a government official, an insurer, or a financial institution, if you’re trying to assess risk and predict social and financial returns on infrastructure investments, you’re running into at least three broad classes of problems…
Complexity, in that communities are complex systems. Any halfway-realistic model will include hundreds of variables, and thousands of causal pathways joining them. A system that large, with that many interdependencies, is going to be beyond the capacity of any human brain to comprehend—and yet that’s what almost all municipal decision-making is largely based on: human beings trying to model community systems mentally.
Even if you could model the system, there are going to be missing data. And noisy data. And the fact that external drivers like the climate and political environment are themselves at best probabilistic.
For any given location, the climate will be changing differently along different dimensions. For many locations, spring onset will be coming earlier—if you look at temperature. If you look at rain patterns, it’ll be coming later. And some crop phenologies won’t change at all, despite the environmental changes. So the climates we expect to see, for the most part, don’t currently exist. And in our world fo judging future performance from past performance—in everything from potential employees to potential investments, we have very little
That last point—the absence of reliable, historically tested best practices—is what drove us to develop the Elephant Builder. The EB is a modeling tool, and the modeling is done by large groups of lightly trained stakeholders and community members. The EB analyzes traces out long causal paths to identify root causes of problems, root drivers of resilience—and it can make policy recommendations. Depending on how you build the model, it can also identify potential systemic failure points—and (this is a brand-new feature) it lets you parameterize the model with quantitative causal relationships, and test quantitative scenarios.
Our method of addressing complexity is to embrace it. We think you should model as much of the system as you can, and if that means hundreds of variables then
But just with just nodes and connections, you can go a long way toward understanding cross-sector interdependencies, and problems and solutions that span sectoral boundaries. The EB will recommend actions, and all the recommendations come from your project’s or your community’s system model.
If you choose to parameterize your model with actual equations and numbers, the EB lets you analyze your model as a Bayesian network. Bayesian networks are excellent at handling uncertainty. They accept probabilistic inputs, and their probabilistic outputs plug right into financial and risk models—and all in units that make sense to human beings.
And all of this is based on models that are built node by node by stakeholders through a user-friendly, mostly unintimidating process.