During his time as a postdoctoral fellow at the Santa Fe Institute, Laurent Hébert-Dufresne, now an assistant professor of computer science at the University of Vermont's Complex Systems Center, began looking at how biological viruses interact with informational memes. Hébert-Dufresne co-authored a February Nature Physics paper on this topic with fellow former SFI postdoc Sam Scarpino and Jean-Gabriel Young from the University of Michigan. SFR spoke with Hébert-Dufresne about this work and its possible application to COVID-19 via phone last week. The interview has been edited for space and clarity.

Some of the research for your recent paper started in 2015 when you were at SFI. What sparked your interest?

If you remember, late 2014 was the peak of the Ebola outbreak in West Africa. A lot of the coverage [in North America] would focus on funeral practices in Sierra Leone and how these cultural traditions were a challenge to the [World Health Organization] in fighting this outbreak. I really got fascinated in how [the way] we talk about public health crises influences the crisis itself, but also the messaging itself is an integral part of this public health data. Social communication around outbreaks are just as valuable data points as cases and diagnoses from doctors. The measles outbreak in the Philippines also is a really good example, sparked by an anti-vaccination movement. I think a lot of what's going on with coronavirus is going to become yet another example of how social messaging and social communication and social contagion are integral to forecasting emerging outbreaks.

What would be helpful for people to understand about exponential growth as it relates to outbreaks?

Exponential growth is the idea if patient zero of an outbreak infects two people and those infect two people each and those four human cases infect two people each, those numbers grow very, very, very quickly. So even though it might look like three cases in a big state doesn't matter that much, it doesn't take long for those three cases to turn to seven and from seven to turn to close to 20, and within days you're at hundreds. What exponential growth is telling us is you can't really rely on a number to estimate what your risks of getting the disease tomorrow are. The risks are growing very, very quickly, versus something like car crashes, where essentially your risks tomorrow in your car are pretty much the same as today. You can't use the same logic to account for risks as a linear process like car crashes. There is more and more evidence that some outbreaks can grow even faster than exponential growth, so we start thinking about super exponential growth.

…One way to think about super exponential growth is that it's exponential growth but where the rate of transmission keeps going up. So, maybe in the first generation of transmission, I infect two people and they each infect two people, but eventually they start infecting three people each, four people each, where the rate is going up, and then we need to think about risks in just a completely different way…

Are you arguing that’s what’s happening with COVID-19?

I don't think it is, based on the number we're seeing. It was a concern for me early on just due to the amount of misinformation or people who were convinced that it was just another flu.

You said on Twitter that messaging/viral stories/misinformation are shaping #COVID19. What were you referring to?

The first example is the official intervention in Wuhan at first downplayed the outbreak quite a bit. It was messages on social media that went viral and told the government to open temporary shelters that over a few days housed tens of thousands of individuals, so it was made bottom up; it was people making sure that the public was aware of the risk that forced the hand of the government into taking the situation more seriously. We're seeing something similar in the US where, in the initial days, it was really fascinating and powerful to me to see how people started social distancing and started talking about the importance of isolation and social distancing before any top-down intervention to close restaurants or close schools and all that. So, I think in a lot of good ways social media has been helping shape this outbreak for the best. We're going to see… where some people are not taking the disease as seriously as they should and then that's going to have the opposite effect.

You’re quoted in Science Daily saying ‘[T]en friends telling you to go see the new Star Wars movie is different from one friend telling you the same thing ten times.’ Is that a way of understanding simple versus complex contagions?

Yes. Exactly. So, in the social sciences when we think about contagion, and here we might be thinking about whether or not to adapt social distancing, the same thing goes: If you have 10 friends telling you to not go out on Friday night, that's more powerful than a single friend telling you the same thing 10 times. This idea of social reinforcement doesn't happen in diseases. To give you a gross example: Ten friends sneezing in your face one time each is not that different in terms of exposure to the virus than one friend sneezing 10 times in your face. The classic picture of all epidemiological models is that all that matters are the total exposures to a virus or the pathogens and that holds true for a lot of diseases. It turns out that complex contagion models work very well as a description in terms of interacting diseases.

[Using] the example of influenza and pneumonia, if you have both, the pneumonia makes you cough quite a bit and maybe that makes influenza even more transmissible. That means you're more likely to infect people if you have both than if you just had one of them. This interactions across pathogens looks a lot like a social contagion. We're at this turning point in mathematical modeling…where we're trying to go beyond one disease equals one pathogen and acknowledge the interaction across different pathogens. We're trying to get new tools to do it better, and I think borrowing from what the social sciences have been doing is going to be a key steppingstone.

Is COVID-19 complex or is it too soon to tell?

It's probably too soon to tell, but the classic exponential growth is holding true and holding true in a lot of different continents, states and regions. What's interesting to me is to see the different rate of exponential growth in different states, some of which are due to population densities, some of which are due to social behavior. Trying to understand what shapes the exponential rate is a complex question. I think a lot of it has to do with social messaging and has to do with us and not the pathogen itself, obviously.

Social media is a relatively new phenomena and contagious diseases are not. Does that make the research you’re doing only relevant in the here and now?

What we're talking about has been always true. Social messaging and the way people talk about outbreaks have always been important in shaping those outbreaks. The reason we're sort of at a turning point now is because social messaging reaches further than before and we hear about outbreaks way before they hit us, and we hear about them not only from official health agencies but from people all over the world. It's important now more than ever to account for social messaging around outbreaks, but for the first time we also have the data to help us do so.

What are the open questions you’re looking at now?

Right now, one key focus of my work and a lot of people at SFI is trying to figure out the next steps. The early work was focused on how likely is this to become a pandemic? And now for the last few weeks, it was more about messaging, making sure people were aware of the risk. The next step for questions is: When is it safe to come out? When is it safe to relax social distancing? That's a very hard question and that's the next big open problem. We need a lot of different perspectives to think about that well because there are still consequences for both health and economics, it's a tough problem but I'm confident that we'll do good work.