In Part 2 of this article, I described why it’s particularly challenging for people to take the right actions in the face of the coronavirus. Now I’ll show you the solution.
“…The big entity that isn’t in the room that would help to quench a pandemic is technology…”
— Tara O’Toole, CEO, In-Q-Tel
A simple decision model for this situation is shown below:
A model like this helps you decide whether and when to start self-isolating and encourage your friends to do likewise. This model is greatly simplified to be easily explainable; we’ll talk below about how to make the model more realistic. But first here’s how it works. You, the user (1) move a lever on the screen representing the date you’ll begin self-isolation. In response, an animation (2) shows the resulting pathogen spread. You can then (3) move a lever to change the number of your friends you can convince to do the same as you. Simultaneously, animation (4) moves to show viral transmission of good decisions/actions, and the (2) animation updates in response. Behind the scenes, a simulation model uses (5) up-to-date data. And (6) incidence curves warp to show your impact, and (7) bars on a map grow and shrink indicating geographical spread. Final outcome animations (8, 9, 10) also move.
This model is crowdsourced (11); you can select different expert teams and also click to a community “model wiki” to read and/or question the justification for the model.
An important note: it is not necessary that these models be perfect, rather that they help us so that we can make better choices than we get when assembling the model elements in our heads (or worse, just doing what others do). And transparency of reasoning in a curated site is key so that we can surface and avoid hidden agendas. See this book for more benefits of this technology.
Another note: the blue triangles represent where machine learning/AI or other types of models inform the chain of events from action to outcome. This is the simple and intuitive way that the power of AI is integrated into an advisory system for human decision-making.
The model above is so simple that you might ask why we need it at all? The answer is that it fills the most important gap in coronavirus decision making by enhancing and aligning intuition is how actions lead to outcomes. Although this is useful on its own, a qualitative model like this can be extended to include more levers, outcomes, and external factors like the effect of weather on virus transmission.
In addition, the computer can help us to answer questions like, “What is the optimal mix of actions that minimizes the harm done by the virus, the harm done to the economy, and the harm done by unintended consequences of the actions?” What is the optimal path to avoid 2.2 million people dead, 20% unemployment, and the effect of social isolation on 2.9 million food-insecure seniors and 30 million children who depend on meals at school? Where can we leverage AI technology and medical and epidemiological technology to understand and quantify how actions lead to outcomes?
This dynamic simulation is immediate and visceral, and something like it should be used by the media, individuals, policy makers, and business leaders like me to understand and communicate the impact of decisions in complex systems. It must be easy to create, and not require coding, to open this understanding to a wide range of people, not just professional modelers. And because these models are innately intuitive, there are many citizen science and open source opportunities to contribute.
The bottom line is that not only is the coronavirus a novel pathogen, but its behavior is also exponential, interdependent, and context-sensitive. We must match this challenge (and the inevitable similar ones) with new breakthrough tools that make knock-on effects, unintended consequences, and the effect of individual systems across time and space visible, intuitive, compelling, and agile, with our own new, democratized, and crowdsourced understanding of complex systems.
“…without careful thought, intervention in a complex situation will, more often than not, lead to unintended consequences. If we don’t lift our game, I promise you, we’ll be left stumbling in the dark.”
—Mark Zangari, writing in The Decision Command Fallacy: How to Avoid Unintended Consequences of Decisions in Complex Environments
A version of this post originally appeared on my Link blog.