The Network Science of “Shelter-in-Place” for Coronavirus

After several weeks of the Coronavirus craze, I of course turned to network science for a fresh look at where we stand with the COVID-19 virus.

Here in San Francisco, we have been ordered to “shelter-in-place” starting today. How did we get here?

The network science view of a virus spreading

Network science examines the ties between things. In this case, things are people, and ties are people’s potential exposure to other people, who might have a virus.

To cope with this virus, we started with large organizations mandating that people work from home. In the context of network science, this was a reasonable strategy because of the nature of what we call hubs and long ties. Hubs are places where people interact and gather, and come from all over to do so. And then, importantly, leave. Leave is the key word. (More on that later.)

So most (all!?) viruses require a few things, network-wise, to sustain themselves. While they vary in their particular needs and persuasions, here are two main concepts that enable a virus to spread:

  1. First, viruses like high density. The easiest way to describe density is simply how “close” people are in a given place. A large office consists of hundreds of people working together, often without physical walls between desks, for example.

  2. Second, most viruses like what we call long ties. Short ties are those between you and your spouse, or child. Quite frequently, short ties are geographically short, too. You live together, for example. Long ties are the opposite. They traverse the network, both in terms of reaching new people, and spanning longer geographic distances. 

Viruses need density to get a foothold in a population, and that’s where hubs come into play. The closeness of ties in a hub such as an office allows a given virus to jump quickly from person to person. 

In turn, though, viruses need to grow and expand — you know, up the ante. If a virus just stays within a given hub, everyone gets infected, and the virus is trapped. Often, pending the strain, the virus effectively burns out. That is why particularly deadly and fast acting viruses can burn out quickly — because no one is left to carry them. (Watch the Vox Explained “Pandemic” episode on Netflix for a great little summary of past viruses that have done just that.)

Leaving a hub

So as I mentioned before, leaving the hub is the main reason why hubs are the problem. Corporations recommended their employees to stay home at the start of the Coronavirus pandemic, not because they were concerned that all of their employees would get each other sick, but because of the extreme exponential impact if a single sick person entered that building, and then the entire building, at the end of the day, left that building.

Once people in a given hub are infected, the virus benefits most from them leaving said hub. Here is where long ties come back into play. The virus jumps on these vectors to new people, and places. And this process is additionally exacerbated if the virus doesn’t cause immediate and/or visual symptoms, and has a long incubation time. 

Now, to make matters even worse, we often go to places via hubs, or go from hub to hub. Airport to airport, train station to work, work to gym. And every time we leave one of those hubs, we give a little helping hand to the virus.

So, while government bodies try to assess the best way to lower the curve of the virus while keeping some semblance of normal life around, the best thing to do is eliminate these hubs. Hence, cities around the country (and world, of course) mandating for the closure of gyms, restaurants, bars, yoga studios, hair salons, etc. etc. Remove the hubs, shorten the ties, lower the curve.

Why did it start with working from home?

Because most networks of people are actually low in density, but connected by hubs, removing a single hub is hugely effective in breaking a network.

Sounds great, right? But if you’re anything like my wife, who was not satiated by this, and wanted to know if there was a network science perspective on why there would be mandated working from home while coffeeshops stayed open and people simply brought their work to the coffeeshop hub instead of the office hub??

…Here’s the deal.

Work is qualitatively different (hub-wise) than a coffeshop (or gym or grocery store), in that exposure time at work is far higher. If you bump into someone once at Whole Foods, and they have a virus with a 10% chance of transmission, you have a 1 in 10 chance of getting said virus. However, if you bump into that same person ten times at work because you CANNOT STOP talking about The Bachelorette, you now are effectively 100% likely to receive the virus vector. And it gets worse!

Coupled to this are the qualities and behaviors associated with work. Unlike the gym or solo work at a coffeeshop, most of us engage in high-density social settings in the service of work. We seek out our work friends, participate in highly engaging meetings, eat lunch in a shared cafeteria, collaborate with team members, and perhaps sit in a shared open area with an entire floor of colleagues. For 8 hours. Yeeeeesh.

Oh yeah, and then we hop on a train.

So — for now, we are sheltering in place. We’ll see in several days if we are doing a good enough job of it to lower the curve.