The total impact of the coronavirus pandemic—the loss of life and the economic, social, and psychological costs arising from both the pandemic itself and the policies implemented to prevent its spread—defy any characterization. Though the pandemic continues to unsettle, disrupt, and challenge communities, we might take a moment to appreciate and applaud the diversity, breadth, and scope of our responses—from individual actions to national policies—and even more important, to reflect on how they will produce a post–Covid-19 world far better than the world that preceded it.
In this brief essay, I describe how our adaptive responses to the coronavirus will lead to beneficial policy innovations. I do so from the perspective of a many-model thinker.1New York: Basic Books, 2018More Info → By that I mean that I will use several formal models to theoretically elucidate the potential pathways to creating a better world. I offer this with the intent that it instills optimism that our current efforts to confront this tragic and difficult challenge will do more than combat the virus now and teach us how to combat future viruses. They will, in the long run, result in an enormous number of innovations in policy, business practices, and our daily lives.
Different model types
The models I describe are not complicated. Each model includes only a few details intended to capture key aspects or dimensions. That approach results in a lower dimensional artificial world—a model—within which we can apply logic, draw inferences, explore, predict, and design.“We live in fragments, disconnected from people, work, and community, our energies focused on social distancing to reduce the spread of cases and fatalities.”
The first model conceptualizes the pandemic as a set of constraints. The pandemic alters the physics of our social, political, and economic lives. We live in fragments, disconnected from people, work, and community, our energies focused on social distancing to reduce the spread of cases and fatalities. Some activities can be accomplished virtually, but others cannot. And, among the latter, those deemed essential, such as medical providers, police officers, and providers of public utilities, have been obliged to continue operations with new constraints: masks, social distancing, smaller teams, and so on.
Added constraints, by definition, reduce the set of possible actions and strategies. First-order logic implies that productivity and efficiency should fall, and costs rise. We are being asked to tie our shoes with one hand bound behind our backs. We assume that we cannot be as effective as we used to be, but that may not be true. To see why, we need other models as I now show.“The investment model implies that the pandemic has made some infrastructure projects cost effective.”
We can embed the constraint model within a model of infrastructure investment—for example, giving each school-age child a laptop and high-speed internet along with access to educational software and more, an enormous outlay of resources. Pre-pandemic, the costs may have outweighed the benefits. During the pandemic, that inequality might reverse. Once purchased, the laptops and software, if not the internet connections, represent sunk costs. Once bought, we reap benefits for years. The same logic applies to building bike lanes in cities and expanding high-speed internet. The investment model implies that the pandemic has made some infrastructure projects cost effective. We will reap the benefits for years.
Another model, the multi-armed bandit model, suggests that innovation will occur because previously risky actions become worth trying. The bandit model represents potential policies and activities as arms on a slot machine. Business or government usually corresponds to an arm with known, and relatively good, payoff.
In normal times, we don’t pull the other arms because they’re not thought worth the risk. Yes, they could be better, but it’s not likely. The coronavirus prevents our use of the current arm. So, we try others. When we pull those other arms—when we allow some people to work virtually or when local governments implement home electrical inspections—we may find those arms to be better than the one we pulled for decades. We may find that we like green eggs (policies) and ham. Were it not for the coronavirus, we never would have tried them.
For example, commuting to work in the District of Columbia when there is more than an inch of snow can be a nightmare. We have now learned that many people can work virtually at 90 percent efficiency; obliging those people to spend 30 percent of their day getting to work is inefficient. It won’t happen in the future. We pulled the other arm. We won’t go back.“Government policies, like business strategies and operational protocols, consist of multiple dimensions or attributes.”
The multi-armed bandit model presumes that we know the set of alternatives, represented by the bandit’s arms. That generally will not be true. Government policies, like business strategies and operational protocols, consist of multiple dimensions or attributes. A fourth model, the rugged landscape model, represents a policy or product as a point in multidimensional space. The landscape model represents the value or efficiency of policy as an altitude. Policy construction consists of searching for points of high elevation—peaks on the landscape.
Interdependencies and the vast scope of possibilities imply that competition and innovation need not necessarily result in locating the global peak on that landscape. More likely, we find ourselves at local peaks. Using this model, we see the coronavirus not as imposing constraints, but as an earthquake that has raised and lowered regions of the landscape. Our previous policies and practices likely no longer occupy peaks on the landscape.
The landscape model suggests at least four implications. First, we should, and do, see small movements from existing policies—making some work virtual, mailing ballots, etc.—to find a peak near the previous one. Second, we should see some long leaps, such as restricting a portion of city streets to bikes. Third, we should see copying of successful leaps. This will happen both from organizations and governments copying similar entities.2Cambridge University Press, 2018More Info → Toronto may copy New York and vice versa, and, if past is prologue, from smaller entities borrowing ideas from larger ones.3Charles R. Shipan and Craig Volden, “The Mechanisms of Policy Diffusion,” American Journal of Political Science 52, no. 4 (2008): 840–857.4 London, Ontario, may well mimic leaps from the London across the pond.
Fourth, returning to the core insight from the multi-armed bandit model, some of these new policies may prove better than what had been done in the past. They will stick even after a vaccine has been introduced or herd immunity reached.“Under Covid-19, governments have incentives to search for new ways to meet constituent needs.”
Another model, the search model, builds on the landscape model by assuming that innovations require searches for technological solutions. Any search has a cost and a benefit. Search occurs if the benefit exceeds the costs. In developing the iPhone, a touchscreen keypad was seen (correctly) as crucial to occupying a very high peak on the smartphone landscape. That realization drove search and led to a key innovation—letters that enlarge upon touch. Under Covid-19, governments have incentives to search for new ways to meet constituent needs. Innovations will follow.
The next model, a spin glass model, considers ensembles or vectors of policies. Imagine the collection of all activity in some domain as being performed either in virtual (V) or physical (P) space arranged in a network. In a simple spin glass model, the value or efficiency of performing an activity either virtually or physically depends on the proportion of its neighboring activities performed the same way. These are known as coordination effects.
As a metaphor, think of each V as a magnet with a positive spin and each P as a magnet with a negative spin. Electromagnet forces imply that each magnet will align with its neighbors. For example, if the activities are arranged in a line, an activity performed virtually surrounded by activities performed physically, as shown below, would be out of alignment. In economic terms, it would be inefficient.
For example, brainstorming virtually using Slack or some other software may be less efficient if related activities like strategy setting and presentation preparation occur in physical space. But, if those activities go virtual, as shown below, then brainstorming as well might be better done virtually.
This simple spin glass model has two equilibria—all physical and all virtual. If we make a more complex network and if we assume that activities have intrinsic advantages from being performed virtually or physically in addition to the coordination effects, then there may be multiple equilibria in which some communities of the network operate virtually and others physically.
This model, borrowed from physics, implies that we may see tipping points. If sufficient numbers of activities go virtual in response to the pandemic, then related activities may then go virtual to coordinate. We may well find that some of these clusters of activities are better performed virtually yet had never found those clusters before because doing so required switching multiple policies or actions simultaneously.
Our last model links policies to culture and culture back to policies. The efficacy of any policy depends in part on the culture within which it operates. We know from cross-cultural experiments and policy failures that incentives and information operate differently in different places. Peoples’ beliefs, their levels of trust, their behavior repertoires, and their social networks all contribute to the success or failure of policies.
The expansion of our behavioral repertoires during the pandemic creates opportunities for future policy innovations. Consider two domains: democratic participation and education. The potential more online interactions with governments, not just electronic voting, but virtual feedback and deliberation, could become the new status quo. Could one have imagined members of a small community discussing spending priorities participating in Zoom breakout groups six months ago? Can one imagine them not, six months from now?“Governments, businesses, nonprofits, communities, and families will reap the benefits of those skills in the years ahead.”
Online education has been moved forward decades due to investments in infrastructure (the sunk cost model), which in turn have driven the development of new software. This fall, professors will use Loom to enliven PowerPoint slides. Students will learn to read documents deeply as part of a community with Perusall, to build idea communities using Slack, and to create presentations on the fly with Sway. The skills that enable students to thrive in those spaces will translate to other domains. Governments, businesses, nonprofits, communities, and families will reap the benefits of those skills in the years ahead.
In closing, the pandemic has had tragic consequences. And it has obliged governments, organizations, and families to respond to new constraints and solve new problems. The innovative solutions that have enabled us to survive this crisis will be a foundation for an even better world in the future.