Michael L. Barnett is an assistant professor of health policy and management at the Harvard T.H. Chan School of Public Health and a primary care physician at Brigham and Women’s Hospital. Caroline O. Buckee is an associate professor of epidemiology at the Harvard Chan School. Yonatan H. Grad is an assistant professor of immunology and infectious diseases at the Harvard Chan School.

As Americans face another four weeks of social distancing, it’s understandable to ask: When can we stop living this way? We don’t know. And we should prepare for the possibility that we’ll be staying inside even beyond the Trump administration’s new April 30 timeline. A date isn’t the only thing that’s unclear: We don’t yet know how social distancing should end, either.

We have to use the valuable time this shutdown buys us to learn what works. If we take advantage of it, we can plan a return to normalcy that prevents covid-19 from roaring back and minimizes the economic costs of future slowdowns.

One of the challenges to determining a date is that there are dozens of separate social distancing policies now in place, and they were implemented at different points in the pandemic. These rules cover everything from school closures to early release for prisoners. The United States isn’t shutting down in tandem. We aren’t all on the same timeline.

But one unexpected benefit to the haphazard, staggered way that social distancing rolled out across the United States is that it created a host of natural experiments.

Some states, such as California, implemented strict policies to limit gatherings starting March 19. Others, such as North Carolina, waited until March 30. Still others, such as Florida, have no such policies, even though their number of covid-19 cases is increasing.

Comparing the outcomes in each state can teach us which policies had their intended effect. Did people stay at home, whether in response to government orders or of their own free will? And did getting people to stay put produce the result we all want: a decrease in infections and the hospitalizations that might follow?

We also need to measure more than just whether early lockdowns worked and whether citizens are following government orders. Different parts of society might be able to start back up at different times. For example, we still don’t know the extent to which children contribute to transmission, so it’s not yet obvious how much school closures help. If we find that children — who often have few symptoms — don’t tend to infect other people and that school closures did not tamp down infections, we could open schools earlier and get relief for parents pulling double duty.

Another question is whether closing restaurants and bars has a meaningful impact on rates of infection. Right now, we suspect that taking the economic hit is worth it in terms of lives saved. But a clearer sense of that calculation will be enormously important for small businesses across the country.

Getting rigorous evidence from these natural experiments will require massive, real-time data collection. We need as much information as we can get on infections and hospitalizations from public health departments, hospitals and clinics. But we also need anonymized data from cellphones, social media and domestic travel to measure whether people actually stayed put. Researchers have already combined data such as this to analyze the impact of China’s distancing policies in Wuhan. Efforts to gather data on mobility in the United States and examine similar questions are also underway.

This isn’t just an academic exercise. Our epidemiological models point to the possibility of another outbreak in the fall. Social distancing keeps people from getting infected, but it also means that many people will still be vulnerable to covid-19 when they are allowed to venture out of their homes. So even after we get through this first wave, we likely have another one coming, and maybe more after that.

This means that we could again face this set of choices about what to lock down, when and for how long. And we might have to make these decisions over and over until we have an effective, widely distributed vaccine, a large fraction of the population has been infected, or we develop other interventions to manage the disease and its spread.

Figuring out the answers to all of these questions is especially important to protect our most vulnerable individuals and communities, including older adults in nursing homes and the millions of Americans with underlying health issues. There will likely be high-risk communities that do not experience much infection, and therefore do not have many immune people to prevent the disease from taking off. This could lead to a disaster as distancing policies wind down if not done carefully.

If we can combine data on covid-19 cases and the effectiveness of specific social distancing policies, we can develop clear guidelines on what and when to lock down, and how and when to emerge again. Staying home can flatten the curve and prevent our health-care system from being overrun. But it can also buy us the time we need to learn how to keep our communities both safe and functional.

Read more: