Ocean Currents

How Do Health Officials Figure Out How to “Flatten the Curve”?

The science behind those social simulations and why they’re critical for good policy

US Capitol
© Thomas Hawk

With the entire D.C.-Maryland-Virginia area now under stay-at-home orders, our team at Ocean Conservancy is fortunate to be able to continue our conservation work by teleworking. You are all in our thoughts as we navigate these new and concerning waters.

Besides “social distancing” you may be increasingly hearing the term “flattening the curve.” Nearly half of all Americans are living the reality of “flattening the curve”: making individual choices in our daily lives that we hope collectively will slow the spread of Covid-19 in our communities. Implementing these public health measures was a big deal—how did our public officials decide that these were the “right” policies to follow?

We’ve learned, of course, from the experiences of other countries. But another major input on these decisions have been simulation models that help predict the effectiveness of different policies. For example, a study from Imperial College London made headlines a couple weeks ago for its influence in prompting United States and United Kingdom governments to take action around closing schools, bars, restaurants and offices in the last week.

A widely shared, simple simulator from the Washington Post provides a good example of how this type of model works—looking at the effectiveness of a range of responses, from a no-action “Free-for-all” to extensive and effective social distancing (see image below). While these models range from extremely complex to relatively simple, they’re both individual-based models.

Screen Shot 2020-04-01 at 2.41.48 PM
© The Washington Post

An agent- (or individual-) based model is a tool that simulates the interactions of individuals within a population. By modeling individuals (instead of populations a whole), we can understand how individuals will respond to a set of circumstances, rather than assuming the whole population will respond in the same way. The seemingly unimportant behavior of individuals can be amplified through multiple interactions to become major drivers of a system. We’re collectively learning this right now as we begin to understand the importance of our individual choices on social distancing in order to help slow the aggregate pace of the disease in our communities.

© The Washington Post

This is important in epidemiology for a couple of reasons. First, differences between individuals change how disease spreads. For example, some people are more susceptible to disease than others (e.g., older individuals and people with preexisting conditions). Also, different kinds of households connect differently and may increase or reduce risks. For example, if there are more extended families that put old people and young children in contact, the economic impact of school closures would be less costly, but it also increases the spread of germs from small children to a more vulnerable population. Second, during epidemics, understanding individual-to-individual interactions are important for policy. Contact tracing, for example, requires a model that can keep track of who meets who, and simulate the ability of the government to follow those connections to develop treatment plans, quarantines and policies.

Social simulation models like these allow policy-makers to ask critical “what-if” questions and try out different policies in a virtual world. This helps policy makers identify good solutions and avoid unintended consequences. This is especially important in moments of great change when policy-makers are asked to make decisions in completely novel circumstances.

Ocean Conservancy’s fisheries team started looking into using agent-based models for conservation about four years ago as part of an effort to improve fisheries sustainability. We developed the POSEIDON model, essentially a “flight simulator” for fisheries management, to support decision-making as ocean conditions begin to rapidly change with climate change, and to help design new management systems in places with large fisheries and little to no history of management.

The other big reason why we looked to agent-based models is that managing fisheries is about managing fishermen, not just fish. And, as we’ve seen in the effectiveness of public health policies, individual choices and the different incentives/risks individuals face can really matter in whether a policy is implemented effectively and has its intended result.

In POSEIDON, we built a model based on not only the fish but all of the components that make up a fishery, including social and economic considerations, collecting data through surveys with fishermen. As you can imagine, each fishery across the world is different, and they are all complex. We know small scale fishermen probably won’t respond to management in the same way as the captain of a big commercial boat because their incentives, expectations and boats are just all-together different. This changes how policies are going to play out over space and time, and we think that matters a lot for the future of good, science-based fisheries management.

Here’s an example of a simulation from our work with the deepwater snapper fishery in Indonesia:


But whether we are determining ways to manage fisheries or manage the spread of a virus, agent-based modeling (built on accurate data) can provide us with a sense of what actions we need to take to keep our ocean, our planet and ourselves healthy.

Learn more at about the future of global fisheries.

Related Articles

Back to Top Up Arrow