COVID-19 and Complex Systems

COVID-19 and Complex Systems

In various mailing lists about the COVID-19 pandemic, I’ve seen several discussions of “complex systems theory” as, possibly, a way to understand how the pandemic is playing out in different locations. Specifically: why have Japan and Hong Kong not experienced an explosion in cases, even though their governments responded poorly to the crisis? The argument is that some systems are intrinsically difficult to model.  There are too many causes and too many effects that interact with each other in ways that are difficult to predict or even understand.

What does that mean? “Complex systems” has always implied, at least to me, systems of nonlinear differential equations, chaotic processes, and the like. Think about the motions of a double pendulum. I wouldn’t deny that there’s a complex set of mathematical relationships behind the pandemic’s spread. But the point of the discussion is that we don’t know those relationships, and discovering them is a slow and difficult process.

Learn faster. Dig deeper. See farther.

As attractive as it may be to think about complex systems, I’m not convinced that’s the best approach. I’ve always been attracted to explanations that are based on a combination of small effects, which themselves are (to a first approximation) randomly distributed. That seems like a much more useful and fruitful way to think about the results we’re seeing in different countries.

What if we thought like this: There are a large number of factors that may or may not have an effect. For the pandemic, that list of factors may look like this:

  • Wearing masks
  • Social isolation
  • HVAC (heating, ventilation, and air conditioning)
  • Social cohesion; importance of the social group relative to the individual
  • Effective healthcare system
  • Temperature
  • Exposure to sunlight
  • The language you speak (Japanese evidently requires less exhaling)
  • Air pollution
  • Obesity
  • Diabetes
  • Heart disease
  • Quality of reporting
  • Quality of testing
  • Genetic factors
  • Virulence of different virus strains
  • Housing density
  • Congestion in public transportation systems
  • Citizens’ ability to self-organize

You get the idea; this list could easily go on, possibly including hundreds of factors influencing disease propagation, individual susceptibility, and mortality. Some are important; some are no doubt fake science; some are just weird; and some may be no more than educated guesses.

We can think of these factors as randomly distributed, although they may not be in reality. In some locales they will add up to minimize spread, and in others, the reverse will happen. You can’t control for, or even measure, several of these factors. But, since we’re thinking of them as random, the fact that Hong Kong has had a relatively minor outbreak that appears to have been controlled quickly means that several factors that limit spread were present, and added up.  The next step is figuring out what the factors appear to be common in best- and worst-case situations.

So, if we look at Japan, we see a society where it’s normal for sick people to wear masks; it’s common decency.  Wearing masks as a prophylactic measure isn’t the big cultural leap that it has been in the United States. We also see a society where obesity is relatively rare, particularly compared to the US (the most obese nation in the world). And Japan is a country in which healthcare is free to all citizens—again unlike the US, where a serious disease can easily lead to bankruptcy, particularly with unemployment (and the numbers of uninsured) skyrocketing. The point isn’t that masks, hospitals, or anything else is some sort of a “silver bullet” that makes the problem go away, though in the US we’ve no doubt undervalued all of these things. Nor can you fix obesity or diabetes or air conditioning as quickly as you can hand out masks. But if each of these factors makes a small contribution—and if, some places, all of those small contributions line up—they have a large effect. Together, they can easily be much more important than a government’s response. And we might start to suspect that some of these “small” factors, such as wearing masks, aren’t so small after all.

Hong Kong is a particularly fascinating case. As Zeynep Tufekci argues in The Atlantic, Hong Kong’s citizens were able to self-organize a response, despite their government. That ability no doubt stems from the unrest and demonstrations of the past year. And self-organization no doubt stems from a sense of social cohesion and responsibility that the US has largely abandoned, if it ever had it. (In some future pandemic, might we see that effect protecting communities that are currently protesting police brutality in the US?  It’s worth noting that, at least in protests I’ve seen, an astonishing number of people were wearing facemasks. And that there hasn’t been a surge in COVID-19 cases that can be attributed to the protests.)

What’s most interesting is that the ability to self-organize probably wouldn’t have been on many lists prior to Tufekci’s article. In college, a professor I worked with told me about his frustration with a defense project he was working on. He was asked to evaluate the probability of an attack on a nuclear facility. His reaction was “we can evaluate all the scenarios on your list, and the probability will be a very small number. But all that tells me is that, if you’ve left a scenario off your list, it’s likely to swamp everything that’s on it.  It’s easy to beat a probability of .00001.” Black swans 40 years ago. Except in Hong Kong’s case, they’re not black; they’re golden.

Which brings us back to the point: understanding the COVID-19 pandemic is less about complex systems than about understanding randomly distributed factors, including some that may surprise us. When we see a group of factors that lead to a good result, we know what to do: hand out masks, provide healthcare, promote the idea that people are responsible for each other, enable spontaneous action, and (to mitigate the next pandemic) wean a society off of its dependence on sugar.

At this point, we’ve made observations and learned how to act effectively. It is increasingly clear that wearing masks is a key to shutting down a pandemic; and that self-organization can play a decisive role to governmental opposition to enforcing mask-wearing. That isn’t to say that we shouldn’t understand the other points: the influence of HVAC, the virulence of different pathogens, the response of different immune systems, and so on. Nor is it to say that we shouldn’t try to understand the system in all its complexity.  That may be necessary to develop effective treatment. But you don’t need to understand the complex system to develop an effective response to a crisis.

Now take a step back and think about applying these ideas to other issues.  What about the re-emergence of racism and neo-nazisim in the US?  Just to be clear—they were never gone.  But, just as with COVID-19, we need to understand the forces that will suppress outbreaks.  Racism has been manipulated for hundreds of years, and is deeply embedded in many of our institutions.  As with any disease, what are the forces that drive it, and what are the actions that disable it?

What about economics?  As the increasing gap between the wealthy and everyone else drives our economies to the brink, what are the forces—the many small forces—that drive us back from the edge?  Viewing economics as a complex system of differential equations is actually an oversimplification.  Can we observe the many factors that line up (and in many countries, have lined up) to give people voices and choices about their future?

For COVID-19 and all the problems we’re facing, that’s the real work, the hard work that can’t wait for the modelling to be done.  What are the many causes that will provide a push in the right direction? We don’t need definitive answers before taking steps.  Once we take those first steps–whether it’s wearing masks or demilitarizing police forces or providing universal healthcare–it will become more evident what works, what doesn’t,  and what the next steps are. And we might discover that effective solutions aren’t as difficult or as distant as they seem.

Source of Article