Science in The Time of COVID: Michael Lachmann & Sam Scarpino on Lessons from The Pandemic

Episode Notes

COVID-19 hasn’t just disrupted the “normal” of everyone’s social practices in what we take for granted as “daily life.” The pandemic has also, more granularly, changed the way scientists research and publish; it has changed the way science interfaces with institutions as varied as local governments and cell phone companies; it has changed the way we host and produce this podcast. This episode, for instance, with SFI External Professor Sam Scarpino and Resident Professor Michael Lachmann was recorded live over a year-end Donor Appreciation Zoom call, for those who both contributed to SFI in 2020 and could handle yet one more group video chat. In it, we discuss their lessons from the “front lines” of network epidemiology this year: what has surprised them, what has stayed with them, and what they expect it all to mean in the years to come…

Welcome to COMPLEXITY, the official podcast of the Santa Fe Institute. I’m your host, Michael Garfield, and every other week we’ll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers developing new frameworks to explain the deepest mysteries of the universe.

Tis the season, so if you value our research and communication efforts, please consider making a donation at — and/or rating and reviewing us at Apple Podcasts. Avid readers take note that the SFI Press’ latest, Complexity Economics, is now available as a free ebook with donation at You can find numerous other ways to engage with us at — and undergrads, you still have until January 11th to submit for our 2021 Undergraduate Complexity Research program at Thank you for listening!

Join our Facebook discussion group to meet like minds and talk about each episode.

Podcast theme music by Mitch Mignano.

Follow us on social media:
Twitter • YouTube • Facebook • Instagram • LinkedIn

More Resources:

Michael Lachmann’s Google Scholar Page

Sam Scarpino’s Google Scholar Page

The University of Texas COVID-19 Modeling Consortium Public Dashboard

Crowding and the shape of COVID-19 epidemics

SFI’s Twitter thread re: auto-correlation in networks on the cusp of a breakdown or breakthrough

Asymptomatic transmission and the resurgence of Bordetella pertussis

Sam Scarpino on Complexity Podcast Episode 25

Harvard’s Michael Mina (hosted by Michael Lachmann) at SFI speaking on rapid testing for COVID-19

“How Data Became One of the Most Powerful Tools to Fight an Epidemic” by Steven Johnson for The NY Times


“If Cancer Were Easy, Every Cell Would Do It” (SFI Press Release on Lachmann’s cancer research)

Episode Transcription

The original machine-generated transcript was produced by Human-edited version courtesy of Caroline Siegel at SFI. If you would like to volunteer helping to edit our podcast transcripts, please email michaelgarfield[at]santafe[dot]edu. Thank you and enjoy:


Sam Scarpino (0s):

What's important for my perspective, is that getting experience working in and valuing interdisciplinary teams. And certainly a part of that is the kinds of things that happen at Santa Fe, where you just spend time immersed with a lot of diverse researchers and you understand more and more how to operate in that kind of context. But then I think also having the interest and humility to work and learn as a part of those groups is really important. So I actually really hope that we can kind of shift the messaging a bit to kind of talk about how important it is that we have these interdisciplinary collaborations, but that people come to the table with an interest to help and learn and, and be productive as opposed to, you know, just try to make statements and take attention and resources and energy away from individuals who are working and have that expertise.


Michael Garfield (1m 8s):

COVID-19, hasn't just disrupted the normal of everyone's social practices in what we take for granted as daily life. The pandemic has also more granularly changed the way scientists research and publish. It has changed the way science interfaces with institutions as varied as local governments and cell phone companies. It has changed the way we host produced this podcast. This episode, for instance, with SFI External Professor Sam Scarpino and Resident Professor Michael Lachmann was recorded live over a year end donor appreciation zoom call for those who both contributed to SFI in 2020 and could handle yet one more group of video chat.


Michael Garfield (1m 50s):

In it we discuss their lessons from the front lines of network epidemiology this year. What has surprised them? What has stayed with them and what they expect it all to mean in the years to come. Welcome to complexity the official podcast at the Santa Fe Institute. I'm your host, Michael Garfield. And every other week, we bring you with us for far ranging conversations with our worldwide network of rigorous researchers, developing new frameworks, to explain the deepest mysteries of the universe. This will be our last episode of 2020. We'll see you again in the middle of January 2021, TIS the season. So if you value our research and communication efforts, please consider making a slash podcast, give and/or rating and reviewing us at Apple podcasts.


Michael Garfield (2m 41s):

Avid readers, take note that the SFI press’ latest, Complexity Economics, is now available as a free ebook with donation at You can find numerous other ways to engage with us at Santa, and undergrads, you still have until January 11 to submit for our 2021 undergraduate complexity research program at slash UCR. Thanks for listening and happy holidays.


Alanna Faust (3m 10s):

Hello to everyone. For those of you that don't know me, my name is Alana. I'm the one who's been sending you many emails. And we often at the end of the year, have some sort of in-person winter tea to thank all of the people who support SFI and support this research. We cannot be in person this year. So much of what we do is now virtual. And I was really excited that Michael Garfield was willing to move the Complexity podcast and have it live and let all of us spectate and then participate. Yeah, I mean, this is a big, thank you to all of you who allow us to do our jobs and I hope you enjoy yourself.


Alanna Faust (3m 59s):

This is recorded. So everyone's going to be muted for the duration. After the conversation, if you have a question, there will be a section for Q and A, and you can either raise your hand or put your question in the chat. If you do have a question in the initial part during the conversation piece, you're welcome to put that in the zoom chat and Michael may see it and include it in the conversation. Otherwise you can wait for the later portion. There's also closed captioning available at the bottom of your screen, and you should just be able to toggle that on and off.


Alanna Faust (4m 39s):

So I'm going to introduce the people who are spot lit right now. First off we have Michael Lachmann. Michael Lachmann is a theoretical biologist whose primary interest lie in understanding evolutionary processes and their origin. His work focuses on the interface between evolution and information. Lachmann received his Bachelor's of Science at the Tel Aviv University, and then his PhD in Biology at Stanford. Both of our guests today were actually post-docs at SFI. Then Lachmann went on to work at the Max Planck Institute and is now a Resident Professor at SFI.


Alanna Faust (5m 22s):

Our second guest Sam Scarpino is an Assistant Professor in Network Science at Northeastern University, as well as an External Professor at SFI. Scarpino’s research spans a broad range of topics and complex systems and network science, including forecasting and predictive modeling, complex network analysis, epidemiology, genomics and transcript genomics, social networks, and decision making under uncertainty. Scarpino’s research on COVID-19, Ebola, whooping cough, and influenza has been covered by the New York Times, the Washington Post, NPR, Vice, Bloomberg, Stat News and many others. Scarpino earned a PhD in Ecology, Evolution and behavior from the University of Texas at Austin, and as I mentioned, was an Omidyar Fellow, a post-doc at SFI.


Alanna Faust (6m 9s):

And lastly, we have Michael Garfield, the wonderful host of the Complexity podcast. Michael Garfield studied Ecology and Evolutionary Biology as an undergrad at the University of Kansas, and spent seven summers doing paleo-ecological field work for the Wyoming Dinosaur Society. And after 13 years as a scientific illustrator and transmedia artist, he joined SFI in 2018 to translate Complexity Science for social media and post this very podcast. So I'm really thankful that you're all here, and I can't wait to listen to you talk about what has been a rather eventful year. Thank you.


Michael Garfield (6m 57s):

Excellent. Yeah. So I think the right place to start with both of you gentlemen would be to just talk a little bit about, I mean, it's already the case that every researcher at SFI has very diverse interests and is their hands in a lot of different research questions, but it's nonetheless the case that for both of you, as well as for many others in this community, this year has precipitated a rather significant pivot in your research activities, and in your relationships to other scientists and to other institutions. And I know in your case, in particular, Michael, you've kind of farmed yourself out to the University of Texas COVID modeling consortium this year with external professor Lauren Ansell Myers.


Michael Garfield (7m 45s):

I'd love to hear from you first and then from Sam, how this pandemic has changed the way that you practice as a scientist and how it's changed the kind of research collaborations and the kind of consulting work and advisory work that you've been involved in. That seems like a good place to start.


Michael Lachmann (8m 10s):

Yeah, I mean, my work since say end of March changed totally. So I, I offered myself to Lauren to do anything she wants in her consortium. And since then I've been there as a programmer and modeler and I do it all the time. So I totally dropped all my other projects and work on it from like 20, 20 hours a day. And slowly, I mean, at the beginning, I have a mathematical background that covers modeling of epidemiology, but I've never done it slowly as I worked on it. I became proficient and trusted myself to do more things. But yeah, the work now is very different from what I did at any time before.


Michael Garfield (8m 54s):

How about you, Sam?


Sam Scarpino (8m 56s):

I think from my perspective, it's been similar in some ways to past experiences with Ebola and with H1N1, although for a much longer period of time and a much more organizational way, especially in terms of how it's affected my life outside of practicing research. So I just finished teaching 150 person intro stats course remotely, which is not something I necessarily expect that I'd ever be doing at least at least synchronously. I think for me, the two biggest changes have been one, how much I interface with journalists and in what capacity. So in the past, I've definitely talked with journalism regularly about my science and work that our group is doing.


Sam Scarpino (9m 35s):

And since the pandemic started, that's happened a couple of times, but primarily it's talking with journalists who are just interested in understanding more about a particular aspect of COVID not necessarily something that I'm directly working on. And so just the types of conversations and the volume of conversations is completely different than something I've never experienced before. And similarly, I spent a great deal of time working with city governments, mayors, city councils, their emergency response teams, doing similar work, trying to help them understand as much as I do or at least tell them the things that I don't understand about COVID 19. On the research side, and it's actually kind of interesting, I'd be curious to hear Michael's thoughts on this.


Sam Scarpino (10m 17s):

And so most of my research group does not work on infectious diseases and they've all contributed to COVID in a variety of different ways, but I have been pushing them to turn back to their own research projects because they have dissertations to complete and research projects that they're continuing to work on. So trying to strike that balance between their desire and interest to contribute, but also wanting to make sure that they finish on time and that the research project they're excited about and that they've poured their energy into continue to progress. That's been something I've been trying to balance as well.


Michael Garfield (10m 51s):

Yes. So listening to you speak about the stuff I am called back to David Kenny's contribution to our Transmission essay series back in what seems like the Dawn Age of March, 2020 when he made it really interesting point in that piece about how scientists prefer to carefully disclaim their research findings, to talk about uncertainty, to talk about margin of error, but when it comes to advisement on public policy, our political actors are public leaders. Heads of institutions wants clear, actionable advice. And in our email lead up to this conversation, both of you addressed challenges in this particular case with different kinds of the poverty of data and radical uncertainty that we've been dealing with in this situation.


Michael Garfield (11m 45s):

So Sam, when we had you on the show back in episode 25, you spoke to some writing that you did on how this is linked specifically to economic inequality and the poverty of data coming from disenfranchised populations. How that affects the way that we respond to this. Michael, you've spoken about your interest in rapid testing, which is an area where we're called to take less accurate, less precise results. Michael Mina gave a great talk on this. That's up on our YouTube channel about how sampling at a higher rate with fuzzier results can give us better information and a better opportunity to respond.


Michael Garfield (12m 27s):

So I'm curious just where both of you are on this issue of uncertainty and actually how to turn into that particular problem, and how to possibly to use uncertainty in our favor in this situation, or where it's still a crippling issue for response in this particular case and in the address of other complex issues, predicaments.


Michael Lachmann (12m 53s):

Yeah. In our work where we try to model Austin or Texas, there's big problems. So the problem here is that we don't really choose our data. We just get data, and very often we can’t control the quality of the data. So when we get data about hospitals, it's hard for the hospitals to provide it. They need to enter all the patients every day exactly. And then, and then it can happen that say one hospital, didn’t do it one day. And if we don't know about it, or even if we do know about it, it makes modeling much, much harder. So the additional noise that is in the data that we can't control because everything is crazy makes modeling, and because of this also predictions, much harder to do.


Sam Scarpino (13m 40s):

Yeah. I mean on the data side, this is one of the things that we got involved in right away, back in January, it it's kind of become the standard operating procedure for infectious disease outbreaks that academics will start essentially hand coding data as it gets reported across myriad sources. So you can remember back to January, we were reading news reports about cases that were happening outside of China, cases in China, and that most of the information around the pandemic was actually things that were reported via journalists because the regular reporting systems from the countries, the WHL, et cetera, hadn't been set up yet. So I was a part of this volunteer consortium, about a hundred people from all over the world that were manually entering in data.


Sam Scarpino (14m 22s):

As it was reported in the newspapers, we entered tens of thousands of individual-level anonymized COVID cases by hand moving into February. And then we spent the last six months working with a team of software engineers who were on loan from Google to build a cloud data platform that automated the process of pulling in all of these disparate data sources, aligning them, cleaning them, ensuring that they're interoperable doing two, a D duplication process. And so we've now got about 7 million individual-level COVID cases, 160 countries. And this is what the New York Times Magazine Stephen Johnson covered when he said that this is perhaps the most complete portrait of this pandemic it's collected anywhere.


Sam Scarpino (15m 3s):

And these data have been used the power all kinds of scientific studies all over the world. If you look at the, the chats, the real-time affective reproductive number, they use the data that we captured from China on the delay distribution between symptom onset and PCR confirmation on how that shifts in time to calibrate their model estimates that everybody are looking at. So we kind of attack the data problem from trying to go to the source and work on that. And now we've got, we are actually very fortunate to receive funding support from a couple of large foundations, it hasn't been announced yet, to continue this work going forward, both for COVID, but then also for future pandemics to ensure that we have the data systems in place that people like Michael and Lauren's team in Texas rely on for their predictive modeling.


Michael Garfield (15m 49s):

So like a related question, you brought up, you kind of alluded to a moment ago, and both of you addressed this in the lead up email thread, which is the way that this situation and other crises create all hands on deck opportunities that tend to pull people into collaborations. That would seem at least on the first pass outside of their area of expertise. I mean, this is perhaps just a microcosm of the way that a rapidly changing world requires a rapid response and challenges are established metrics for expertise in the first place, but it calls up all of this question.


Michael Garfield (16m 32s):

Like Caroline Buck, you talked about this when she was on the show, about people that are swerving out of their lane, and offering advice on issues about which they know nothing, and they're blocking the communications channels and increasing the noise in the discourse around this, making it hard for people who really ought to be given the microphone and opportunity to speak. This is an especially difficult problem for female scientists and scientists of color. It's aggravating existing social inequities, but at the same time, it kind of speaks to this kind of broader issue with Complex Systems science and interdisciplinary research in the first place.


Michael Garfield (17m 14s):

And I'd love to hear you talk to both the issue of Sam, what you called epistemic trespass. I love that phrase and where it is and is not an appropriate accusation of what's going on here. And you know, where it is that people without formal training in epidemiology or a related area really are needed in this. Because I think the way that research collaborations are taking place now and the way that this has accelerated the sharing of on peer reviewed preprints and so on is a window into the way that science is going to be practiced more commonly in the years to come.


Michael Garfield (17m 56s):

And so these problems seem really worth attending to now. I'd love to hear your thoughts on that. Whoever cares to speak first.


Sam Scarpino (18m 3s):

Well, I'll be brief cause I'm definitely curious to hear more about Michael's experience with Lauren and her team, but, you know, I am, I'm very torn about this because I think we've seen a lot of high-profile examples of deeply unproductive statements coming from individuals that don't have the expertise and experience to be making them that are sucking up the communication channels and the attention as you said, but I also don't have any formal training in epidemiology. And I think one of the interesting things that happened is that a colleague of mine, Professor Brandon Bruno who's at Yale, wrote this article for Wired magazine called “COVID-19 Carpetbaggers” and how to identify them. And he listed me as an epidemiologist.


Sam Scarpino (18m 44s):

And then I tweeted and said that I'm not an epidemiologist. And then Twitter was arguing about whether or not I was an epidemiologist. Nobody could agree. And so I think one of the things that the real challenge is like on the call right now, the three of us should be talking about how spiral shaped molecules cells evolve or something, right? Like not about COVID like our background is in seemingly unrelated fields. Although of course, as Michael mentioned, the math that that describes how populations change can be applied to lots of different systems. And so I think what's important for my perspective is that getting experienced working in and valuing interdisciplinary teams and certainly a part of that is the kinds of things that happen at Santa Fe, where you just spend time immersed with a lot of diverse researchers and you understand more and more how to operate in that kind of context.


Sam Scarpino (19m 26s):

But then I think also having the interest and humility to work and learn as a part of those groups is really important. So I actually really hope that we can kind of shift the messaging a bit to kind of talk about how important it is that we have these interdisciplinary collaborations, but that people come to the table with an interest to help and learn and, and be productive as opposed to, you know, just try to make statements and take attention and resources and energy away from individuals who are working and have that expertise.


Michael Garfield (19m 55s):

And that's a constant issue in our Facebook group, reminding people to operate with negative capability, come in, eager to learn. Yeah. Michael, what about you?


Michael Lachmann (20m 5s):

Yeah, Lauren's group, so there's like the UT Consortium. It is really amazing how many, how diverse the people who work there are and also how many people offer their expertise. So there are mathematicians, physicists, engineers, all just want to help [inaudible] asks all the time, ask whatever, you know, like, is there any coding that they can help us with? So that group is really amazing. And how diversity is people who just want to come and help. The same is actually also true for the rapid testing group, like rapid We're also there's people who just want to help.


Michael Lachmann (20m 47s):

And they say, you know, like some know video editing, other know how to talk to people, things like that. So I think that that is really amazing in this pandemic. In terms of expertise, it’s a very complex issue because I think that everybody who works on COVID is way over the work. I mean, it is a, like I said, you work all the time and any help would be good, especially the COVID epidemic is I think very complex. And these days it's not the virus anymore that we need to model. Actually it has been from the beginning, but now even more, it's not the virus that we need to model. It's the people. it's hard to understand what the people will be doing.


Michael Lachmann (21m 29s):

And I don't think that we don't really necessarily have the right expertise to do it. So I think that any help should be accepted with open arms, but of course, on the other side, it's not easy to talk in a way that will help. It took me several months until I understood the language enough so that I could contribute. And I know friends of mine who wanted to help model wrote a very nice model, but the problem is it's really hard to get it into the community. The community knows to language that they're talking in and the model that comes from some other fields is not understandable. So I think it's a problem. Like on one hand, you really need a lot of help, but on the other hand, it's really hard to give you that help.


Michael Garfield (22m 13s):

So I'd like to peg from this into a question that ties into research that you've both done on human social relationships, because I think one of the things that we've continued to bang the pan about at SFI --  David Krakauer brought this up very early in our Transmission series -- was how this particular epidemic and epidemics in general, in so far as complexity researchers like to think in terms of flows of information, then the flow of the information contained in a viral genome is kind of fungible in some ways with other kinds of information that flow through human social networks.


Michael Garfield (22m 59s):

And, Sam, you had a paper that you did with Laurent Hebert-Dufresne and Jean-Gabriel Young on macroscopic patterns of interacting contagions, how they're indistinguishable from social reinforcement. Michael you've done work on costly signaling and the body's inhibition of cancer signals to try and cheat their way into more nutrition from the body. So, you know, I think about all of this and how it's related to how the lowered barrier to access in discourse through the erosion of legitimacy and traditional institutions, and new sources, has allowed more bad actors into the conversation has escalated, informational warfare, the proliferation of QAnon this year to be piggy-backing on the epidemic.


Michael Garfield (23m 49s):

And I'm just curious how you two feel that your research in this area does or does not eliminate this issue. And you know, what it might mean for how we think about systems of human communication in the future to not only inhibit the spread of future epidemics, but also future epidemics of socially deleterious human behavior and human belief. If we can go there.


Sam Scarpino (24m 19s):

You want to go first, Michael, or


Michael Lachmann (24m 21s):

I can try. I mean, I think that there is a question, what is driving changes in the epidemic? So for example, in the early modeling of, in Austin, you can see right there, was an early time when the epidemic was really growing exponentially and then it stopped. And it was around the time when a lot of mitigation strategies were enacted, like lockdown, and schools closed and so on. But I think what we see in our data is that the numbers started to go down in Austin before lockdown. It started to go down when schools closed and college closed.


Michael Lachmann (25m 2s):

Even though today, we think that directly schools don't really cause so much outbreak. But I think what happened was that the closing of schools was a signal to people that this is really serious. So they started to respond already. And we also see it later, where if the governor of Texas announced that people should wear masks. And I think that was again a very strong signal in itself, even if people didn't start wearing masks immediately, but they treated the disease more carefully, more seriously. So I think that a lot of the measures aren't just measures in what they do. They're also a signal to people what is happening with the disease.


Sam Scarpino (25m 41s):

Yeah. I think that it's a really interesting and challenging question from a whole bunch of different directions. So the paper that you mentioned, Michael, we were showing that a classic model that a lot of social scientists use to study information sharing on networks, the social reinforcement, where it's kind of like a voter model for how information transmits. So you're going to continue to hold the belief you have, unless both Michael and I are saying otherwise, and then you'll switch your belief as opposed to just getting infected from me only. So there's kind of this threshold. And what we show is that those are analytically equivalent to models that biologists use to study interacting pathogens, moving through this, the same population of posts.


Sam Scarpino (26m 21s):

And why am I telling this story? Well, I think the interesting part of this to me is that the co-authors of the paper and I actually disagree about what the most important take home message is. So they think the most important take home message is that we can take a single time series and use our social reinforcement model to ask whether there's any evidence for interacting pathogens or any other kind of weird non-linearity that's present when normally it would take at least two times series or, or more data to do that. And so basically this like silver lining, what I see is, well, we actually don't have a way of inferring the mechanism that's causing something to spread from the time series data alone, that you have to bring in external information and think about how you actually evaluate mechanism from a lot of different angles.


Sam Scarpino (27m 6s):

And I both find that to be perhaps pessimistic, but also very exciting from a scientific perspective that it means we have to bring to bear lots of different models and data and inference methods and expertise to try to understand what's going on. And to Michael's point raises a really interesting question, really, we saw the same thing in Boston. It was one of the reasons why I and others were so critical of the IHMI forecasting models is that they said, okay, if you have a shelter in place mandate, here's what happens to transmission. And that's not what happened. People lock themselves down weeks before any of that went into place because they were scared. And so the real thing that we have to understand is what is influencing people's behavior. How does that interact with their social networks and the pathogens to end up affecting or not a disease transmission?


Sam Scarpino (27m 49s):

And so even from the very, very earliest days of COVID, which should have been about as simple the model as it's ever going to get like a fully susceptible population, almost literally one person walks off of an airplane into the population that immediately becomes vastly more complex. As soon as people start to respond to that pathogen in terms of how they are going about their day-to-day lives. And I think that's all the more reason why we need the kinds of interdisciplinary collaborations that we started this conversation with.


Michael Garfield (28m 17s):

So to that point, you know, looking at network structure, whether it's in collaboration networks or whether it's in the structure of our cities and how they're organized, I mean, it's clear that what counts as an effective intervention in one Metro area doesn't seem to work across the board, that we have to think about this in terms of how different cities have grown up over time. And I know Sam, you just coauthored a paper in Nature about this, about the metapopulation structure of our cities. I'd love to hear you give us a short exegesis of that work. And maybe if we have insights, the two of you can draw from that into thinking about how, you know, just for example, like the, I remember Richard Florida published a really interesting piece a couple months ago on his blog about how the structure of our cities has already reacted to previous epidemics that, you know, the porcelain bathroom, wasn't a thing until we realized that germs were lingering and wooden toilets and that kind of thing.


Michael Garfield (29m 26s):

And so we've the way that we have come up with the design of our urban spaces has changed a lot. And, you know, other than the apparent and horrifying emergent norm of, you know, constant face masks and the new radius of personal space that we might just be sort of left with after this, I'm curious about these insights and how you imagine they might change the way that we react to one another going forward in the 21st century.


Sam Scarpino (29m 53s):

Yeah, well, I'm really fascinated about that in general. I think one of the things that very few of us have appreciated is how much trauma we've gone through. Even if you haven't had COVID or don't know anyone directly who's had it, or worse that it's affected our lives so dramatically that we're going to be processing this for the rest of our lives. And in the same way that, you know, I remember my grandfather telling my dad before he left to go to college, how you could try to talk food out of a restaurant if you're hungry. And it's clearly, you know, his memory of growing up in the depression and going hungry and that this has this organizational effect on the rest of your life. And I think we're all going to be doing that in ways that we haven't really come to understand yet.


Sam Scarpino (30m 34s):

And I suspect a lot of that will affect our cities and how they're organized. And so one of the things we tried to do in this paper is answer, or at least start to try and peel back some of what Michael was mentioning around why this epidemic persists in some places and seems to burn out quickly and other places. And it turns out that there's quite a bit of literature on this, both in the theoretical and empirical sides of epidemiology. So papers from Duncan Watson, Peter Dodds, looking at the effect of hierarchical community structure on an academic process. And so what you kind of imagine is you've got households that are in neighborhoods that are in larger areas of cities, and then cities have these kinds of conglomerates of these neighborhoods, and that depending on the biology of the pathogen, you may have like these little mini outbreaks that are happening in households, and then it takes a little bit longer to jump into another household in a different neighborhood. And so if you actually had the really high resolution data, you'd see all these little waves that are kind of on top of each other, but because our data are noisy and the system itself is kind of noisy, it's this long kind of smear of cases that just drags on and on and on and on and on. And so we took data from China and Italy that had already completed their first waves of the outbreak and showed that the width and height of the epidemic curves in hundreds and hundreds of cities across Italy and China is well predicted based on a simple measure of population aggregation in space.


Sam Scarpino (31m 60s):

And that this is something that's predicted from theoretical models and actually was shown in one of my favorite papers on epidemiology by professor Lisa Sattenspiel, an anthropological study of fur traders in Canada during the 1918 flu pandemic, where she actually showed that you get this kind of divide between tightly connected groups of fur traders and fur trading outpost versus more loosely connected, and it drives these kinds of multi-wave epidemics and junior coauthors coupled those data like the log book data from the fur traders with the 1918 flu cases with these agent-based studies. So I think that these are the kinds of things that so many of us on the call understand how to work with and model. And that's again, partly why bringing this kind of interdisciplinary collaboration is so important because we do need to learn something about all the ways in which we understand how cities operate and they're organized and how that affects dynamical processes.


Sam Scarpino (32m 51s):

But then as we move forward and we have to think about what do cities 2.0 look like? How do we best keep the things that we think are important about social networks and cities and, you know, the kinds of work that's been done pioneered at Santa Fe Institute, while doing the best that we can to mitigate the future epidemic risks. So I think it's going to be a really critical area of complex systems research and science for years, as a result of our realization about the fragility of so much of our organization, especially in United States and in Western Europe with respect to the effects of COVID.


Michael Garfield (33m 25s):

Michael, this seems related to a topic you brought up ahead of this conversation about the strange behavior of the pandemic, how it, the curve is staying flat much longer than the models would predict. I don't know if you want to link to that or if, but that seems like an interesting association.


Michael Lachmann (33m 41s):

Yeah. I mean, I think that at the middle stage of the pandemic as we saw it in Austin, but I think in many, many other places, that the pandemic has had this long stasis where the cases stayed high, but overall almost constant. And I think in most models that is really hard to get something like this, because most models you will either have, I mean, regular models will have exponential growth or exponential decline, but even if you modify the model to have a network structure, it's still hard to get stasis. You will either have maybe linear growth or gross, like cubic growth or something like this, like to get real stasis I think it's very hard to get to the model.


Michael Lachmann (34m 24s):

And then, so my interpretation was that it has, therefore it has to be, it has to be some kind of feedback. So there has to be feedback that comes from the community where the community sees something about the epidemic and then response. It could even be something as simple as people moving out of from New York, because they don't like it there anymore, that it are so dangerous or people moving to Santa Fe because it looks so risk-free. But I think that Sam could talk about this issue much better than me.


Michael Garfield (34m 57s):

Do you care to, or shall we launch into the Q & A?


Sam Scarpino (35m 1s):

Well, I'll just mention just very briefly that, and I actually think this is something that would be really interesting to work on is we're tracking a lot of mobility data. And one of the things that was true back in the lockdowns in the spring is that every way we slice and dice the mobility data, it was crashing. But now we see that it's coming back up differentially. So for example, over Thanksgiving, we saw nationally that we had near pre-COVID levels of durations of social contacts outside of the household. However, we were still down 50% in terms of the number of unique social contacts. We've seen that mobility outside the household for shopping has come back up to, you know, maybe 60, 70% of normal, but commute flows are still way down, but have been creeping up slowly, slowly, slowly over the past few months.


Sam Scarpino (35m 48s):

And in my head, I can't get this image of the suburbs around Boston, that until the commute flow came back online were previously basically isolated from each other. And then you have individuals going back to work in Boston mixing, and then going back out to the suburbs again. And so I think what we really are seeing is this like really rapid heterogeneity that has evolved in terms of our social contacts and our social networks. And it means that, you know, even trying to answer what seems like a fairly simple question from Michael about why is the epidemic curve so long becomes deeply, deeply complicated. Very quickly.


Michael Garfield (36m 22s):

That question of heterogeneity. We have a question from the audience, Kirsten Canes, I believe. Kevin, if we can get you to unmute Kirsten so she can ask her question to our panelists here.


Kirsten Canes (36m 34s):

Hi, thanks for giving me the opportunity to ask a question. And I appreciate the speakers. This was very informative. I'm interested in racial disproportionalities, and in particular, the evidence that the disease burden is different for different racial groups, but so is the effect of remote learning and the potential risk to miss learning opportunities for people of color. And so I'm wondering how often policy makers and your scientific community asks you to think about those issues. And also what you think are the most promising areas of research to help us make decisions, to mitigate risks for people of color in the United States.


Michael Garfield (37m 17s):

Thank you.


Michael Lachmann (37m 18s):

I can address it a bit. So in our consortium, it's a constant issue that we work with all the time. So our, our modeling of Austin for example, is one big mishmash where all of Austin, other than age groups and risk groups, is considered as one big pot. But when we look at data, we see that it is very much not like that. And we have been since the beginning of the epidemic have been trying to take that into account in everything we do. It is not easy because anytime you try to make the model more precise, it's easy to make the model more precise, but it's very hard to then fit the parameters so that you can use this.


Michael Lachmann (38m 1s):

So we need to have parameters, like how much do people move from one area of the city to another, or how much, how many people work as essential workers and so on. So it's easy to care, but it's hard to actually do something.


Sam Scarpino (38m 16s):

I think just very briefly. I mean, one of the things that we're going to have to figure out in this country going forward is how to prevent this from happening again, if there's another pandemic right. So one of the big challenges that local school boards, mayors, et cetera, we're faced with is that they can't close the restaurants without a federal bailout. Even the States really can't do that. They don't have enough money to bail out the restaurants. They can close the schools and that's doing something. And so I think we saw lots of cases across the U.S. where the schools were closed, because that was the thing that they could do without the federal government support. Similarly, in terms of the CDC, until very recently, they'd been providing almost no guidance whatsoever with respect to what people should be doing.


Sam Scarpino (38m 60s):

And so you have mayors, city councils, school boards that are faced with this decision of trying to decide whether schools are safe or not. And they don't have the data, they don't have the guidance, and they don't have the leadership that they need. And instead, then they fall back on influenza where it is high risk for children, and schools are key drivers. And so they fall back on what they do know, and they close the schools. And we're only now very slowly starting to see that the CDC is providing the kinds of data that as Michael mentioned, shows that schools are well not, no risk, are low risk for transmission, and they can help with the reopening. And of course I'm sure you understand better than I do, and many people on the call do, that schools provide a lot more than just what happens in the classroom with the teacher across the community.


Sam Scarpino (39m 45s):

And so that when the schools are closed, there's a huge cost that the children pay in terms of their development, but behavior, but also in terms of the community and the support that they have. With respect to communities of color, one of the key issues is that quite often the same kinds of effects of racism and xenophobia that caused these higher health burdens also result in a lack of information that's available to public health agencies, to Michael Lachmann and Lauren Ancel Meyers, to actually model what's going on and what the needs are. So one of the ways that governments disempower individuals is by not collecting data on them.


Sam Scarpino (40m 26s):

It's why we were fighting over the census so often. We actually published a paper about this over the summer, showing that individuals who were in the highest risk groups for influenza by socioeconomic status are less likely to be in the surveillance data sites, which means we have more bias in our forecast, less understanding of the demand. And then of course, we move into the vaccine era. You know, we're seeing in Massachusetts that individuals who are non-healthcare essential workers aren't being prioritized for the vaccine. So the same kinds of lack of equity. We also aren't engaging with the fact that many of these communities have earned a deep distrust of the government, especially as it pertains to vaccines and medical care. And we're not really addressing that from a communications perspective, trying to build back that trust.


Sam Scarpino (41m 6s):

And so I think it's a very complicated issue that is going to require a lot of attention to work if we want to be able to move forward at all from where we are right now.


Michael Garfield (41m 17s):

Thanks. So I'd like to open this up, Kevin, if you can unmute Alan Covich. I hope I'm saying that, right. Alan had a question.


Oh, I was just curious about individual based modeling. It seems like that's a pretty widely used kind of thing. And the areas of movement ecology. It seems like it'd be really helpful to short out different risks in terms of how people interact when they're out doing various types of things, be it hiking, bicycling, shopping, whatever. There may not be enough data to really make that helpful. It seems as though it would be a tool that everybody’s been using for lots of different kinds of reasons, and certainly generate the interdisciplinary curiosity that could also develop some new ideas.


Sam Scarpino:

I guess, very briefly, and my [inaudible] has a lot more experience given the kinds of modeling they're doing in Lauren's group, although I think most of it's not agent-based modeling, but yeah, one of the things that's different about COVID is that the technology companies have opened up their data to researchers and public health in a way that they have never before.


Sam Scarpino (42m 20s):

So we have this incredible lens into what individual people are doing as they go about their day-to-day lives. It's actually, I think one of the big issues we're going to have to address as we go forward. So in South Korea, for example, they have this social contract between the citizens and the South Korean CDC, as a result of SARS and MERS that says if these kinds of public health triggers happen, then the mobility data from the telecommunications companies gets handed over to the Korean CDC as a part of the response. So that the public is actually benefiting from the invasion of privacy happening from the telecommunications companies. In the U.S. that's not the case, right? Like, so we have our privacy massively invaded by the tech companies, but then we don't get the data to benefit from it.


Sam Scarpino (43m 0s):

And this is one of the first times when we have actually gotten access to these kinds of data and literally can look and see how people's movements are changing on a zip code by zip code level data by day by day, as they respond to different orders and fear and all of the other things that we've seen. So this is really the first time you had the data to be able to approach, at least the detailed level forecasts with those kinds of models.



Alan Covich:

There was some discussion about having cell phones, being able to interact and actually be able to map locations of individuals that would certainly be some rich data for individual base modeling.


Sam Scarpino:

Yeah, that'd be great. Yeah, absolutely. We were supposed to have all these contact tracing exposure alerts with the Apple, Google collaboration and the Bluetooth and everything else.


Sam Scarpino (43m 47s):

And somebody asked me the other day, what I thought about the ethics of that. And I said, I wish we were talking about the ethics of that because somebody had actually done it instead of, we don't have the damn thing, especially in the U.S., it's just been inaction after inaction after inaction. And so I think we should have had a conversation about whether we would have benefited from that as a society. And if we came down on the side of, yes, we should have done it, but Mike, I don't know what your thoughts are on agent-based modeling. I know that that's an issue in Martin's lab.


Michael Lachmann (44m 11s):

In our group, we do every type of modeling. So the main model we use i is like I said, compartmental, so everybody is just in a group, encountered as a group. But for example, when we model schools, we have individual based models at the various levels. You know, like we have models that look at [inaudible], the students, when they are in the school bus and in the class and walking around, or we have models that just are still individual-based, but then in large batches. And like I said, the group also collaborates with many outside groups. So we have groups of engineers that did really elaborate individual-based models, where they actually do look at individuals on their way to the grocery store and back, and whether they're in the car or a bicycle, but like you said, the problem really is, I mean, there's actually two problems.


Michael Lachmann (45m 1s):

One is in order for that to really help you, you really need to have a well calibrated model. If the model doesn't have the right parameters, it won't give you the right answers. And the other is how fast can it run. So when we do the Texas modeling, the Austin modeling, we need to run the model thousands of times to be able to say, which of the models fits the current, what we currently have observed best, and whereas detailed individual based model it's not possible, or at least in the limited time that we had to write these, we can't make it fast enough. So that is for us, that's the main reason we don't use more elaborate network models and individual based models.


Michael Garfield (45m 43s):

So I'd like to open this up…


Alan Covich (45m 44s):

[inaudible]…a list of things to do, to get ready for the next pandemic though. You know, we were clearly behind, even though people like Bill Gates and others were making these predictions, and I think now everybody's very sensitive to the fact that this won't be the last one. The question is when will be the next one, and how much will we learn. Is there a central place where we're trying to come up with a good list of what we needed to be doing more comprehensively? Or is it still kind of scattered?


Sam Scarpino:

I can be very brief or try to, I mean, this is a great question and we could have a whole, well, we did have some actual SFI meetings on this a while ago and I think it'd be interesting to have one again. I think actually for me, the thing that I'm more worried about is the next 12 months, and many of the things we're going to need to do over the next 12 months are the kinds of things that we're going to want in place for the next pandemic. So for example, we don't know whether the vaccines that are coming block transmission. We don't necessarily have any reason to think that they don't, but that's not how the trials were designed. One of the projects that I started as a postdoc at SFI was to try to answer this question about whether the currently used pertussis vaccines block transmission. We are still working on that question and we are still arguing in different academic camps about what the answer is. That is a very hard question to answer, and we are not designing the right kinds of data collection systems to be able to address that as the vaccines start to roll out.


Sam Scarpino (47m 8s):

As COVID becomes increasingly rare, all of these things that we're talking about are going to get harder and harder and harder and harder. So one of the issues right now in Boston, is that they have to triage out people with respiratory illness that is not COVID. And as COVID becomes less common, the proportion of individuals that show up in the ER in respiratory distress due to something else is going to go up. And so, unless we have high resolution surveillance systems, unless we data sharing, unless we have a plan for how we're going to respond quickly to flare ups, what kind of measures we're going to put in place? It's going to be a mess over the next 12 months. Even if the vaccine blocks transmission at 95%, we still have to vaccinate 60 to 70% of the population in order to get to herd immunity. That's about the percent of the population right now that says they'll get the vaccine.


Sam Scarpino (47m 49s):

And so that's going to take a lot of work, and it's going to take at least 12 months to do it. However, we can be back to a new normal in May, if we layer in non-pharmaceutical interventions, mask wearing, physical distancing, with the vaccines as they come online in a smart way, but we're not planning for that right now. And so instead, we seem to be having this idea of, of some kind of vaccination scrum without knowing whether the transmission is going to be blocked. And even if it was, it's going to take 12 months to get back to a new normal, and we're continuing not to engage in that social contract with the public where we explained to them, here's what we're doing. Here’s when it's going to end. Here's why it's going to work. And here's why we're asking you to make these sacrifices. So hopefully with the administration change, that happens, but we are still going to be months behind by the time that occurs.


Michael Lachmann (48m 32s):

Yeah. I'm worrying about the next two months. I mean the growth that we see now is just incredible and the virus will not stop itself. I mean, we need to do to stop it. It might, if we don't stop it, then the vaccine might become irrelevant. We have to think about it like this, but I think that at all levels, we need to learn about how this pandemic progressed. And I think, I mean, this is out of my pay grade, out of my area, but I think what is to me totally obvious is that everything we deal with now in the next pandemic, we should never have these problems. We should never get to this stage.


Michael Lachmann (49m 13s):

The pandemic, every pandemic should be stopped right at the beginning before it even starts. We should never hear about them. I think there might've been 20 other pandemics that we never heard about because they were stopped at the right at the right time. This for me would be the main thing. We should have stopped it in March or January. But in terms of understanding what we did wrong, I hope we'll have enough time between this pandemic and the next to evaluate that.


Michael Garfield (49m 42s):

Excellent. So I'd like to open the floor to Christian Lemp. We can ask him to unmute and drop his question in here.


Christian Lemp:

Thank you so much. I appreciate it. One thing I was curious about is in this COVID pandemic one massively different dynamic is that we're all connected in a virtual way, much as we are now. And I'm wondering if you know anything about the behaviors that were adopted in previous pandemics, probably at the local community level. And is there anything different now that we're so super connected and we have all these new technologies? Are there any sort of negative outcomes as a result of that?


Sam Scarpino:

Well, I think it's super interesting if for no other reason that I've watched friends of mine become literally celebrities on the internet as a result of their work for COVID-19. It's been pretty amazing to see that happen, especially on the science side of things, and how much effort has gone into trying to combat fake news and fake information by so many scientists that are dedicating a huge amount of time to that on social media and in the news.


Sam Scarpino (50m 45s):

I'll answer this with, to me, what I think is one of the more interesting things I've learned about pandemics from COVID and you guys, I guess you'd think I'd already know this, but it turns out that we don't know as much as we maybe should about what happened in 1918 and how society's changed. I think it's a really interesting question about what happened, why whatever happen didn't seem to last. And actually one of the things that I was reviewing Nicholas Christakis’ book on COVID, is a sentence that says that pandemics are under appreciated in modern literature. That there's not very many pieces of modern literature that covered pandemics. I thought that there's no way that that's true. That can't be true. I mean, he said it in such a way that it was kind of a statement that you, you know, it was unfalsifiable anyways, but I said, no, I can't be true.


Sam Scarpino (51m 28s):

So I started doing some research and it turns out that at least one prominent modern kind of Western cannon author agrees, and that's Virginia Woolf, who wrote an essay about having the 1918 flu. And in that essay, she talks about how literature does not cover infectious diseases the way or, pandemics or anything, the way that you would have thought they were. And she provides some explanation for why she thinks that is. But I think it's a really interesting question for whether that's the case and why, and actually turns out is that we have a last kind of sidebar here is that there are apparently a minority of Wolf scholars that think that the start of Mrs. Dalloway is actually about the end of the 1918 flu pandemic instead of exiting from the Wars to like reveling in the mundane and shopping and those sorts of things.


Sam Scarpino (52m 11s):

And so I think there is probably a lot we can learn. I think there's probably a lot that we've recapitulated that has happened before and has been undone. But I actually think that's a really important question that probably a lot of people on this call would be able to contribute to.


Michael Garfield (52m 24s):

Michael, do you have thoughts on it?


Michael Lachmann (52m 26s):

Not as deep. What I see is mainly from the scientific side, because that is the main thing that I'm exposed to, you know, I'm working in Austin and the same group, there's people who work from LA, from Iowa, and all this wouldn't have been possible. So the collaborations that are possible, it wouldn't have been crossed, but on the other hand maybe if I wasn't working on this, I would work on the origin of life. So I definitely lost that. I say, so I think that the fact that we can in this time interact with our family, even while they are under lockdown, is an amazing contribution of the technology to what we are able to do today.


Michael Lachmann (53m 9s):

But I can't even talk about how much we lost.


Michael Garfield (53m 14s):

So gentlemen, do we have time for one more question here? I guess, as the sort of Wil Wheaton, Wesley Crusher of SFI, this one I've prepared specifically for this call, that if we have time for, I'd like to get to, which is, and we've touched on this already in the call earlier this week in the SFI musicology working group, I saw SFI Fellow Tyler Margaritas speak on his research into creative breakthroughs, you know, findings that right before a phase transition, like the Eureka moment of an individual scientist or the sudden coordinated changes that happen in a jazz ensemble that the networks, whether an individual or collective show a spike in auto correlation. I guess kind of like the hexagonal cells that form at the bottom of a pot of boiling water, suddenly a new structure emerges, and other SFI scholars, like Miguel Fuentes, Raissa de Souza, Martin Schaeffer, have all discussed about how changes in network structure precede, or possibly even help predict things like the frailty of aging brains, the imminent collapse of civilizations.


Michael Garfield (54m 15s):

We talked about this with Jeff West on the show: in spite of social distancing, and as you've discussed on this call, in many ways, we are more connected to one another than we ever have been before. And this is related to the strains of you're talking about everyone being overworked strains on our attention and our time, and our emotional energy. And nowadays every, every breaking news item seems related to everything else. So in a way it looks like the onset of schizophrenia. You know, like even if folks in the complex systems community are quick to celebrate it as an opportunity for it to be a common understanding in the public imagination of the interconnectedness of our world. It seems like we're sort of dipping our toe over the line here as a species between genius and madness.


Michael Garfield (54m 60s):

And so how much is too much? In what ways is more connectivity helpful, and in what ways, other than the obvious ill-advised swapping of bodily fluids, would it be useful to try and preserve weaker coupling in our systems?


Sam Scarpino (55m 15s):

I mean, we could have a whole podcast probably about that. It’s something I've been interested in for a long time, so that we had a workshop or five of us who spent the summer in Marseille. And really, we were just trying to figure how to spend the summer in Marseille. But the question was like, what signal would we see in human social networks of that process happening, right. If this kind of trade-off between the benefits of social connections and the cost of social connections, and especially if we thought it was an out of equilibrium system. And I think it's a really interesting and important area of research that I know many people on this call are working on. And, and so I think those are the kinds of questions that we should be wrestling with as a conflict systems community anyways, but certainly because of COVID because there's going to be increased relevance for what, if anything we can do to try and preserve the things that we think are good about social networks while mitigating costs.


Michael Lachmann (56m 4s):

Yeah. I mean, I think that from what I saw, I'm now all the time just working on COVID, but kind of what I see from there I think the amazing speed of science as it occurred in the COVID research. I mean, this is an amazing breakthrough on one hand, but on the other hand, right, no one has time to on any question deeply. From what I saw in other fields, kind of on the side, I think that other fields, for example, the field of the origin of life, also started to do like accelerated interaction, like accelerated research over this time. And again, we have, I think after we're done with COVID and we're done writing up all the mistakes we made and all the things we need to learn from this.


Michael Lachmann (56m 49s):

I think going over the different scientific fields and understanding how much of an acceleration we observed, how much of a slow down, how has it changed? The process of reviewing and looking at papers is I think a big, a big thing that we'll have to learn from this way of doing science.


Sam Scarpino (57m 7s):

There are something like well over 200,000 papers with COVID in the title since January. And it's interesting because “COVID” is not a word that we used before January. So you can be pretty sure that those were all in the last eight to 10 months. And so it's kind of unimaginable how much information has been generated and put out there by the scientific community.


Michael Garfield (57m 31s):

Well, that's just wonderful. I really want to thank you both for donating your time to this, to everyone who showed up for their donations in a financial contribution or in sweat equity to SFI and everything that goes on here. This episode will go out on our regular feed in a couple of weeks. This week, we're about to publish our conversation with Artemy Kolchinsky, which is really bizarre and wonderful. So I hope that if you're not already subscribed to the show that you do so wherever you go for podcasts and thanks again, everyone so much for facilitating and attending this discussion. This was immensely fun for me.


Michael Garfield (58m 12s):

And I hope for you too, as well.


Alanna Faust (58m 14s):

Yeah. Thank you all. And to reiterate what Michael Garfield put in the chat follow up questions can all, if you're on Twitter, you can tweet about it. You could email me and I can send them on. And thank you for being part of this conversation. Thank you for being part of the SFI community. And I wish you a safe, distanced end of the year -- distanced physically, but not emotionally or intellectually. Bye.


Michael Garfield (58m 48s):

Thank you for listening. Complexity is produced by the Santa Fe Institute, a nonprofit hub for complex systems science located in the high desert of New Mexico. For more information, including transcripts, research links and educational resources, or to support our science and communications efforts, visit