In October 2019, just a few months before a novel coronavirus sparked a deadly pandemic, a group of government officials, business leaders, and academics convened in New York City to role-play a scenario in which a novel coronavirus sparked a deadly pandemic. Their imagined virus leaped from livestock to farmers in Brazil, then spread to Portugal, the United States, and China. Soon, it was everywhere. Eighteen months later, 65 million people were dead.
This simulation, known as Event 201, was one of dozens of so-called pandemic war games run in the two decades leading up to the outbreak of COVID-19. In mid-2020, as the world came to terms with its new pandemic reality, media outlets published a flood of articles about these simulations. Some highlighted their prescience, others their blind spots. But the real-world crisis that occasioned this review was only a few months old. Whatever hindsight it provided wasn’t yet in focus, because many of the greatest challenges of the pandemic—new variants, vaccine hesitancy, the hyper-politicization of public health—were still to come.
Almost three years later, we know that the war-gamers whiffed on many of these longer-term outcomes. Pre-pandemic role-plays successfully predicted early events like the overwhelming of the nation’s hospitals, ineffective travel bans, and a lack of coordination across levels of government. But they underestimated the significance of masking policies, the speed at which vaccines would be developed, and the politicized backlash to those interventions. They also failed to account for cascading viral evolution, and did not grasp how long such a crisis could last. “The scenario ends at the 18-month point,” the makers of Event 201 wrote. “The pandemic is beginning to slow due to the decreasing number of susceptible people. The pandemic will continue at some rate until there is an effective vaccine or until 80–90% of the global population has been exposed.” If only.
War-gamers are trying to learn from their mistakes. Long before any public-health authority had declared the acute phase of the pandemic over, officials were already playing out fresh scenarios that better fit the facts. At the Munich Security Conference in February 2022, for example, the Johns Hopkins Center for Health Security helped run an exercise focused on rapid development and equitable distribution of vaccines for an imagined future virus, Tom Inglesby, the center’s director, told me. Later that year, at an annual meeting convened in Brussels by the WHO and the Bill & Melinda Gates Foundation, the same group ran a simulation in which current and former foreign ministers had to grapple with the threats of misinformation and resistance to non-pharmaceutical interventions. Unlike pre-pandemic war games, this one also took into account governments’ attitudes toward expert public-health advice.
Accounting for the specific facts of the coronavirus pandemic could make future war games more realistic, experts told me, but it won’t ensure that they have substantive value. These exercises are not meant to predict what will happen in the next pandemic, so much as to prepare for what might happen. Bringing games into alignment with recent experience matters only insofar as it conduces to the latter.
War-gamers often draw a distinction between tabletop exercises and functional exercises. In the former, participants sit around discussing what they would do in a given scenario; in the latter, participants actually do it. They might transport stockpiled resources, distribute personal protective equipment, or care for pretend patients at a hospital. The Pentagon moves real forces around the world as part of its simulations. This real-world practice is sorely needed for pandemic scenarios, too, says Jennifer Nuzzo, the director of Brown University’s Pandemic Center. “When I look back at COVID, many of the failings stemmed from the fact that we didn’t functionally exercise enough of the capacities that we supposedly had.”
Pandemic exercises could also be made more challenging, Inglesby said. Too many are set up to prevent participants from feeling demoralized, he said, as opposed to “exercising to the point of failure.” This principle can be taken too far, though, Nuzzo told me. If the exercise is too intense, people get overwhelmed and disengage. Real pandemics, of course, are highly overwhelming, but if you’re trying to train someone to bench-press their weight, you can’t just drop a 200-pound barbell on their chest and shout “Go!” They have to build up strength over time.
Coming through an actual pandemic is a form of training, too, but that experience doesn’t make war games useless. In fact, they may be more important now than ever, experts told me. Before COVID, pandemic role-plays were meant in part to raise awareness—to show participants and the public alike that a pandemic could really happen. Now, obviously, few people need to be persuaded of the risk. Post-COVID, the exercises serve a very different function: They remind us that the next pandemic might look nothing like the one we’ve just experienced. It could have a far higher case-fatality rate. It could disproportionately sicken children rather than the elderly. Its symptoms could be neurological instead of respiratory. “Just having lived through COVID doesn’t prepare us for all future events,” Inglesby said. Experience is an infinitesimal sliver of possibility.
We try, through war games, to internalize this. It is not always easy. Event 201 notwithstanding, pandemic simulations have traditionally focused too much on influenza and not enough on other pathogens, perhaps because of how many outbreaks the former has caused in the past. It would be just as wrong to focus only on coronaviruses now, and leave influenza behind. At the simulation in Brussels, Eric Toner, a senior scholar at the Johns Hopkins Center for Health Security, told me, some participants struggled not to default to recent memory: “They kept coming back to, ‘Well, in COVID we did this,’ or ‘In COVID we did that.’” It’s a delicate balance to strike: to learn from experience without being constrained by it.