BCoSE: an AI-backed planning tool for coastal resilience
A community-built, publicly available, AI-backed tool for predicting and monitoring beneficial resilience outcomes of coastal infrastructure projects.
Infrastructure planning should be a transparent process which takes into account not only direct costs and benefits, but also secondary impacts on community well-being and resilience. But communities are complex—coastal communities exceedingly so. It’s difficult for planners and investors to trace out the myriad intertwined casual paths that connect implementation to social outcomes. For citizens and nonprofits, who lack access to the relevant data and models, it’s impossible.
The Bayesian Community Systems Explorer (BCoSE) is a public outcome-prediction tool. Its back-end is a Bayesian network: a probabilistic network model whose nodes include all the variables that drive and describe climate, seas, ecosystems, livelihoods, infrastructure health, education, justice, cultural vitality and the built environment—and whose connections form the complex web that connects these domains and that form the causal paths from infrastructure to community well-being.
The structure of the model will be built by community-member experts in a series of open modeling sessions. No one will need to choose a most-likely sea level rise scenario or a singular future unemployment rate; instead, BCoSE will accommodate the full range of possibilities and the probabilities of different scenarios. We’ll derive these values from published models and existing data, but also from expert and community-member opinion—a useful feature of Bayesian networks is that their probabilistic parameters accommodate quantitative data and subjective opinion equally well. Users then input whatever numbers they have (observed values, estimates, plausible ranges), and BCoSE calculates probable outcomes for all other nodes, including community well-being and resilience variables—all in units meaningful for decision-making. BCoSE will be available to the community via a web application.
- A planner uses BCoSE to assess the direct and indirect social costs and benefits of proposed infrastructure projects. She uses this information to prioritize and advocate for particular projects.
- An investor uses BCoSE to determine which proposed projects are likely to maintain satisfactory returns under various natural-disaster scenarios. BCoSE produces outputs in meaningful terms, like expected dollars.
- A developer uses BCoSE during project-development to create project specifications. He tests construction and implementation scenarios, checking to see whether the resulting social costs and benefits meet target thresholds. Once he has identified specs that—according to BCoSE—produce the desired outcomes, his company can include those specs with confidence that the government and investors (also using BCoSE) will also find those specs to be satisfactory.
- A community activist is working to shift her organization’s action model from reaction to proaction. She uses BCoSE to identify the consequences of various infrastructure projects and practices, and to build campaigns that are anticipatory and data-driven.
- An insurer uses BCoSE to estimate losses under different climate and disaster scenarios. Like actuarial tables, BCoSE’s predictions are probabilistic. They connect easily to the insurer's models and workflows.
BCoSE is a community-built, quantitative predictive tool, which is deployed as a public web application. A model of the democratization of AI, and of data-driven, resilience-focused, systems-oriented planning, it can be replicated in other locations, ecosystems, biomes, and institutional contexts.
Watch our elevator pitch:
Where our solution team is headquartered or located:Chicago, IL, USA
The dimensions of the Challenge our solution addresses:
What makes our solution innovative:
BCoSE updates our existing tool, the Elephant Builder, which allows communities to build collaborative causal maps of their socioeconomic community systems. Our current customers—city and county sustainability, resilience, and emergency management professionals—are excited about the transparent, inclusive, systems-oriented approach, but want to run quantitative scenarios: "If x, then what y?" BCoSE adds an AI layer to forecast multiple outcomes simultaneously in changing environments. Risk analysis like this has been the domain of high-end consultancies that charge millions just to start the conversation, and used by corporations and insurance companies—but out of the price range of most local governments.
How technology is integral to our solution:
We’re creating BCoSE to support three values, each supported by a technology:
Collaboration: BCoSE uses a network-visualization application to structure collaborative modeling, so the resulting model includes broad, multisector expertise.
Transparency: BCoSE is built on a Bayesian network, an AI technology that allows quantitative, probabilistic inferences to be drawn from causal models that would make sense to most people in a community, leading to agreement about solutions.
Openness: BCoSE is a public web application, allowing any community member with a web connection to access the same information that’s available to the government, leading to better-informed solutions and quicker approvals.
Our solution goals over the next 12 months:
We are seeking a pilot project: a community to partner with us in the creation and deployment of a local version of BCoSE. Over one year, we could work with the community to build and parameterize the model, test the tool for accuracy and usability, and train planners and developers in using BCoSE and interpreting its outputs.
Norfolk would be an excellent pilot community. Members or our team have worked with their CRO in the past, and the entire Hampton Roads region poses useful challenges.
Our vision over the next three to five years to grow and scale our solution to affect the lives of more people:
When BCoSE shows results in the pilot community, likely a place that has been frustrated by seemingly intractable coastal resilience challenges, we will unlock a floodgate of interest—what is this BCoSE, how did she get to solutions that work for all of these stakeholders so fast? Would it work for us? BCoSE can scale up, out and deep to offer solutions to dozens of communities in three years, with at least 100 challenged coastal communities seeing results in 10 years.
The key characteristics of the populations who will benefit from our solution in the next 12 months:
The regions where we will be operating in the next 12 months:
How we will reach and retain our customers or beneficiaries:
Governments or other stakeholders hosting BCoSE will be able to choose how openly to share the tool and which data to make public or keep private, but BCoSE is a web application and is designed to be freely available throughout a community. We would like to see government agencies, private-sector developers, nonprofits, and other stakeholders all drawing from the same model and the same data.
We intend to work with local governments and with nonprofits like ICLEI and 100RC to start BCoSE projects in communities everywhere.
How many people we are currently serving with our solution:
BCoSE’s predecessor, the Elephant Builder, has been used by
48 representatives of water, energy, transportation, communication, health-services, and emergency-management systems in metropolitan Los Angeles to model lifeline interdependencies for California’s Fourth Climate Change Assessment. The outputs benefit local government agencies, utility providers, and the populations they serve.
6 city-government resilience planners in Buenos Aires to build causal models which were used by about 80 stakeholders in a 100RC resilience strategy development process.
8 county-government agencies, businesses, and community groups in Larimer County, Colorado to collaborate in assessing their cross-sector resilience strategy, which will benefit the population of the county.
How many people we will be serving with our solution in the 12 months and the next 3 years:
Demand for the Elephant Builder is growing, and we anticipate expanding to Canada and Finland over the next 12 months. But the Elephant Builder’s evolution into BCoSE—adding quantitative, predictive capabilities—will allow resilience planners not only to facilitate cross-sector collaboration and visualize the whole system, but also to predict outcomes. In 3 years, BCoSE will be helping people in coastal communities around the world to evaluate and choose among alternative scenarios—in ways that acknowledge and leverage the interdependencies and uncertainties inherent in the process.
How our solution team is organized:
How many people work on our solution team:
How many years we have been working on our solution:
The skills our solution team has that will enable us to attract the different resources needed to succeed and make an impact:
Founding member of Global Adaptation and Resilience Investment Work Group, whose members are helping to increase investment in resilience and are looking for investable resilience projects (e.g., equity for a scale out of BCoSE, debt for BCoSE-derived infrastructure solutions).
8,000-member network (Climate Resilience Consulting) of resilience-interested leaders all over the world, including at academic (e.g., MIT, ASU, Notre Dame) and development (World Bank, Climate-KIC) institutions, infrastructure organizations (National Institute for Building Science, Institute for Building Technology and Safety), and foundations (Rockefeller Foundation, Kresge Foundation).
Our revenue model:
Once we’ve developed the core logic layer of the tool, we can customize the city layer for other coastal cities—and we anticipate, based on the record of the existing prototype and on the globally active network of the team, that many coastal cities around the world will be interested. We will sell custom layers to each coastal city, as well as access to the tool and its stored data and consulting and training. Ultimately the coastal-city layer can be adapted to different environments—montane cities, forests, fisheries—with custom layers on top of each.
Why we are applying to Solve:
We have an innovative solution to an intractable problem, and we’re well positioned to implement it—with Solve’s help. We can replace the static, two-dimensional tools of the resilience field, the spreadsheets and impact matrices, with a user-friendly method capable of harnessing complex interdependencies and predicting indirect downstream impacts. Our team has decades of experience in community and infrastructure resilience planning; technological and modeling expertise; and a successful prototype in wide use. Connecting with MIT Solve’s resources and networks would allow us to develop a commercially viable tool that meets the needs of the communities it’s meant to serve.
The key barriers for our solution:
To integrate Bayesian networks into our existing app, we need funding to license a Bayesian network engine, engineers and city-systems experts to help us parameterize our models, and mentorship on running inference algorithms at scale.
To build the initial model, we need a community experiencing shocks and stressors willing to pilot a creative approach, as well as access to local environmental, economic, and infrastructural data and expertise.
To test the solution engine, we need end users—city planners, investment analysts, developers, activists, insurers—to solve for their own interests while seeing the consequences for the rest of the community system.
The types of connections and partnerships we would be most interested in if we became Solvers: