COMPLEXITY: Physics of Life

Reconstructing Ancient Superhighways with Stefani Crabtree and Devin White

Episode Notes

Seventy thousand years ago, humans migrated on foot across the ancient continent of Sahul — the landmass that has since split up into  Australia and New Guinea. Mapping the journeys of these ancient voyagers is no small task: previous efforts to understand prehistoric migrations relied on coarse estimates based on genomic studies or on spotty records of recovered artifacts.

Now, progress in the fields of geographic information system mapping and agent-based modeling can help archaeologists run massive simulations that explore all likely paths across a landscape, bridging the view from orbit with thoughtful models of prehistoric peoples and how they moved through space.

The new research expands our scientific understanding of how ritual and story encode vital geographic features, and sheds light on how our modern world is the product of deep, ancient forces.

Agent-based modeling in archaeology can also help save lives by improving science communication, empowering stakeholders in cultural resource management, and facilitating better international planning and coordination as the climate crisis looms…

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.

This week we talk with Stefani Crabtree, SFI Fellow and Assistant Professor in Socio-Environmental Modeling at Utah State University, and Devin White, R&D Manager for Autonomous Sensing & Perception at Sandia National Laboratories. Stefani and Devin are the first two authors on the recent Nature Human Behaviour paper, Landscape rules predict optimal superhighways for the first peopling of Sahul, a project at the bleeding edge of agent-based modeling for archaeology that simulated over 125 billion potential ancient migratory routes.

In our conversation, we discuss bringing advanced technologies to bear on research into human prehistory; the ways humans make sense of space; how our minds and landscapes inform each other; and the ways agent-based modeling might help avert disaster for the sedentary populations of our century.

If you value our research and communication efforts, please subscribe to Complexity Podcast wherever you prefer to listen, rate and review us at Apple Podcasts, and/or consider making a donation at You can find numerous other ways to engage with us 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

Related Reading & Listening:

• Stefani’s Website

• Devin’s Webpage

• Landscape rules predict optimal superhighways for the first peopling of Sahul by Stefani A. Crabtree, Devin A. White, Corey J. A. Bradshaw, Frédérik Saltré, Alan N. Williams, Robin J. Beaman, Michael I. Bird & Sean Ulm 

• Complexity 60: Andrea Wulf on Alexander von Humboldts Naturegemälde

• Complexity 33: The Future of the Human Climate Niche with Tim Kohler & Marten Scheffer

Subscribe to updates from SFI Press on the upcoming ABM for Archaeology textbook

• Lauren Klein’s SFI Seminar: What is Feminist Data Science?

• Sam Bowles, Wendy Carlin, Suresh Naidu: Core Economics

• Scale and information-processing thresholds in Holocene social evolution by Jaeweon Shin, Michael Holton Price, David H. Wolpert, Hajime Shimao, Brendan Tracey & Timothy A. Kohler 

• The universal visitation law of human mobility by Markus Schläpfer, Lei Dong, Kevin O’Keeffe, Paolo Santi, Michael Szell, Hadrien Salat, Samuel Anklesaria, Mohammad Vazifeh, Carlo Ratti & Geoffrey B. West

• Outreach in Archaeology with Agent-Based Modeling in Advances in Archaeological Practice by Stefani Crabtree, Kathryn Harris, Benjamin Davies, and Iza Romanaowska

Episode Transcription

Machine-generated transcript by edited by Aaron Leventmann:

Devin White (0s): I was able to model fairly successfully big chunks of this ancient trail system that I could actually see from space using satellites. But I would identify ancient trails. I would go out on the landscape, ground truth them, map them out, and then generate theoretical models based on these cost estimate equations and to see if they would line up. It was sort of the early days of the research, but there are these couple of areas where I was working with the archeologist on the range at the time, Adrienne Rankin, who just knows that place inside and out, she's been there a very long time and they kept pointing to this one area. This area doesn't make any sense. Like the trails just aren't doing anything like

I would remotely expect that people are trying to move efficiently. And she said, well, two things. One, you don't understand that landscape as well as the people who've been living there for a very long time. And granted that's a very good point. And she says, also, you maybe need to think a little bit differently about what they are valuing in terms of what are they trying to minimize as costs. I spend enough time out in this region to understand that that area was what archeologists would call a ritual landscape. That the goal moving around in that area was not about efficiently getting from point A to point B.

It was about getting between these very important sacred places. And so it just a completely different kind of movement. So it got me thinking about how do you quantize that? How do you actually get that into a model where it's squishy or so to speak? It's not slope, it's not temperature. It's not time. It's the value of people are putting on reaching a place and that place matters. And it doesn't matter how much it costs you physically to get there. Sometimes making the journey more difficult is the point

Michael Garfield (1m 55s): 70,000 years ago, humans migrated on foot across the ancient continent of Sahul to landmass that has since split up into Australia and New Guinea. Mapping the journeys of these ancient voyagers is no small task previous efforts to understand prehistoric migrations relied on course estimates based on genomic studies were on spotty records of recovered artifacts. Now progress in the fields of geographic information system mapping and agent-based modeling can help archeologists run massive simulations that explore all likely paths across a landscape, bridging the view from orbit, with thoughtful models of prehistoric peoples and how they moved through space.

The new research expands our scientific understanding of how ritual and story and code vital geographic features and sheds light on how our modern world is the product of deep ancient forces. Agent-based modeling and archeology can also help save lives by improving science communication, empowering stakeholders in cultural resource management and facilitating better international planning and coordination as the climate crisis looms.

Welcome to Complexity, the official podcast of 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 week we talk with Stefani Crabtree as a five fellow and assistant professor in socio environmental modeling

Michael Garfield (3m 34s): At Utah State University and Devin White, R and D manager for autonomous sensing and perception at Sandia National Laboratories. Stefani and Devin are the first two authors on the recent nature human behavior paper “Landscape rules predict optimal superhighways for the first peopling of Sahul,” a project at the bleeding edge of agent-based modeling for archeology simulated over 125 billion potential ancient migratory routes.

In our conversation, we discussed bringing advanced technologies to bear on research into human prehistory. The ways humans make sense of space, how our Mayans and landscapes inform each other, and the ways agent-based modeling might help avert disaster for the sedentary populations of our century. If you value our research and communication efforts, please subscribe to complexity podcast wherever you prefer to listen, rate and review us at Apple Podcasts and/or consider making a donation at

You can find numerous other ways to engage with us at Thank you for listening. Stefani Crabtree, Devin White. It is a pleasure to have you on complexity podcast.

Stefani Crabtree (4m 53s): It's very nice to see you and hear you, Michael. 

Michael Garfield (4m 58s): It's been awhile. Stefani. I'm glad to meet you Devin. You two are the first two authors on this extraordinary nature human behavior article on the first peopling of Sahul of what we now think of as like Australia and the related islands. Before we get into that, I want to invite both of you to just give a bit of a background about yourselves as scientists and about the kind of work that brought you into this project, what animated you to become researchers and then like map the pathway from where you started as curious minds into your peopling of this project.


Stefani Crabtree (5m 45s): So my person as a scientist, I did my PhDs at Washington State University and at Université de Franche-Comté. So I have two of them. And while I was doing my PhD in the U.S. my work focus predominantly on the U.S. Southwest and being a Southwestern researcher, it turns out this is one of those lovely places where all the researchers get along and get to know each other, which is a great place to be as an early PhD student.

And I was at our society for American archeology conference, I believe in San Francisco, when I really connected with Dr. Devin White, who at that point was in DC. I believe this was before he moved to Oak Ridge National Labs and his PhD when he was at CU Boulder was looking at human migration within the Southwest, across some really inhospitable landscapes. And I was an agent-based modeler.

That was predominantly what I did. It's kind of like Sim City for researchers where you imbue an individual with characteristics, and you have an environment that changes over time, and you look at what happens and Devin and I decided to meet in the lobby of the conference hotel. And I was late because I had just seen an archeologist get mugged outside of the conference hotel. And I run in, I'm super worried about meeting Devin and here's this person who I've never had like huge conversations with.

And we're like, let's do something awesome together. And so that was really the beginning of our work together was could we figure out a way to bring my expertise in agent-based modeling and lots of people moving across the landscape with Devin's amazing expertise of building this from everywhere to everywhere model that people. His dissertation that really changed the field of how we think about human movement. And so this was many, many years ago before I finished my PhD and we kind of tabled that idea until I kind of settled somewhere and started a post-doc.

And I guess we'll continue that conversation of how we decided on Sahul or Australia and Papa, New Guinea and Tasmania, and the surrounding areas until Devin kind of introduces himself. I'm now a researcher at the Santa Fe Institute and also an assistant professor of socio environmental modeling at the Quinney College of Natural Resources at Utah State University. 

Devin White (8m 24s): That's a tough act to follow. So, but like Stefani I'm with Southwest archeologists, by professional training, she made a lot of great points about why it's such a great laboratory for exploring complexity and prehistory and even new technologies for how to go about doing that. When I was in graduate school, my original focus was very traditional and on Southwest archeology, but there was a professor there named Payson Sheets who got some funding and had been working in Costa Rica for many years on understanding human movement down there, trying to identify ancient trail systems.

But he did it in a very novel way, which was looking for these trail systems using remote sensing. So using satellite technologies that were becoming more available to archeologists at the time. And so everybody wanted to go do this. There was a lot of competition amongst the grad students to get in on that project because you get to spend a summer in Costa Rico. And it's kind of hard to beat that. Now, if anyone's ever done field work in Costa Rica, they'll tell you it's not as glamorous as it sounds, but it is a fantastic opportunity, but he had one requirement. And that was, you had to complete a series of courses within the aerospace engineering program at University of Colorado on remote sensing method and theory, image processing, geospatial technologies, all these really crazy important, very difficult things to do, especially if you're an archeologist and you're not trained in anything like that.

And so several of us went in. Only a few of us survived, but we came out of it inspired and went off and did this work. And I felt like even though Costa Rica wasn’t my area, just trying to understand prehistoric network, how people would move through them, how that sort of shaped their experience, how that helps us understand the past. I could bring that to the Southwest. And so that's what I did for my dissertation was looking at this really inhospitable landscape in Southern Arizona that honestly looks a lot like environmentally parts of Australia that we looked at for the current study.

So there's a lot of overlap there. Most archeologists are all guilty of this. It's just human nature. You tend to focus on smaller areas. And what I think we've seen, Stefani was a pioneer in these generally archeologists that are coming forward that do this now, when we look at larger and larger areas, you know, bigger expanses of space, bigger expanses of time to do that, you really do need to bring technology to bear on the problem. So starting with my dissertation, and then moving beyond that into my professional career, I kept bringing bigger and bigger computational tools to bear on this problem and had to invent a lot of new ways to do so in the process.

What was interesting about our professional meetings, the one where Stefani and I met several years prior to that, I had a very similar encounter with another archeologist when I went to grad school with who had been watching what I had been doing for my dissertation and said, could we take this and apply it to this area of central Mexico to try to understand some really, really important aspects of pre-history down there. And that was really the birth of a new origin of the technology that you applied to this study was I had to take what I was doing for my dissertation and supersize it in this really crazy way. It really set the stage for where we ended up with this study it almost 10 years later.

And so this has been a journey of many, many years, and I've managed through my professional career, which has not been in academia. I've worked in federal agencies in DC and worked at Oak Ridge national laboratory for about six years. And I've been at Sandia for three and a half. Now is this has been a constant theme, trying to understand human movement, patterns of migration. These days it's not my primary job. I actually run our autonomous sensing and perception organization here. So we focus on AI and machine learning object recognition for what we'll call high consequence systems, things where people make decisions based off of these systems that impact human lives.

And so there's a huge trust factor in that in terms of making sure that these technologies are doing the right thing. And so Melanie Mitchell and I think would get along very, very well in that regard. So this is not my day job, but it's something I'm incredibly passionate about. And working with researchers like Stefani and the team from Australia, it keeps me engaged. And it's just been an absolute pleasure.

Michael Garfield (12m 14s): Well, I definitely want to get back to trust issues because there is something about remote sensing and the necessity of satellite data that kind of rhymes with big science and in general, and the abstraction of the research process, and definitely ties into where I want to land this conversation much later in the use of agent-based models as a teaching tool and as a tool for a much more inclusive science and like getting people involved in this stuff.

So there's kind of a Mobius strip here that I want to explore with the two of you around the way that you actually did this research and the way that it requires basically standing on the highest mountain you can, which is in space to get a perspective on navigating this research terrain, but then having to come back down off that mountain. Before we get into the weeds here, I think it would be useful to talk a little bit about the prior research that you referenced in this paper. How have people tried to understand archeological research on migratory patterns in the past? You enumerate a number of different approaches and the limits of these approache.

And I think that that's the right place to stand here to sort of understand the coastline that you and your coauthors expanded with this project.

Stefani Crabtree (13m 38s): So I think it's easiest to begin by talking about when you get a guidebook that tells you about hiking, maybe around Santa Fe or anywhere else, it'll have a trail rating system. And that trail rating system is actually based on how difficult it will be going up, going down, how long the truck is, what you need to carry, if you need to carry water, if there's water along the way. When I was an undergrad in Los Angeles, I was surprised to learn that a lot of trails have like water fountains along the way.

It's just a thing. Those ways of ranking trails are kind of at the basis of what a lot of path models do in a geographic information systems kind of way. But a lot of these GIS approaches take into account your start point and your stop point and what does it take to get between those two? And that's how a lot of models archeological and non-archeological have started from a beginning point to an end point. How costly, how hard will it be?

How much uphill, how much downhill, how much flat, what do you need to those kinds of things? And that's great because it can tell you a lot about how difficult it would be to walk from say, Chaco Canyon to the basin of Mexico, which we know people did. But if you don't know where that path is, it becomes more difficult because you have to give it more and more starting and end points. So we have one-to-one. So that's one path. Now we have two to ones. That's two paths. Well, two to two, that starts getting more and more complex.

Now, 10 to 10, you start doing the math and it starts getting kind of intractable. And then there are a lot of other things that you need to take into account when you do different kinds of models. And so maybe it's not just walking and carrying water, but maybe you want to walk by a certain spot or you want to avoid a certain spot. And so these kinds of things get into the complexity of this. So when we did this analysis, we looked at the landscape of archeological path models and looked at how many paths, places these studies created.

And they were in the tens or dozens. Sometimes if people really wanted to get things going, they were in the hundreds, but the more computational complexity you have, the better computer you need. And it becomes increasingly difficult, ends up out by necessity. A lot of these are pretty simple.

Devin White (16m 17s): There's sort of this siren song when you think about doing path modeling and archeology. I mean, it seems simple. It looks like it could be easy to do. There are push button tools to do it more importantly, we carry this technology around with us everywhere we go on our phones. You know, if you use any kind of GPS based routing to get you from point A to point B, it's effectively the same technology underneath, but since you're running it on your phone, you think, oh, this has gotta be pretty lightweight. Like it can't be that complicated to do this, but almost no one really thinks about every day, including myself is really all the computing is happening in some big, massive data center somewhere.

You just send your request in, and then you get the answer back and you think this is instantaneous, but there's like some massive computing machine in the background that does all of this work. And so it's easy to fall into that trap of thinking, well, I could quite efficiently apply this to some archeological problem. And then you start doing this and you, and you try to do it at a human scale. You know, you think about anything high resolution, like being able to think about our built environment and how much detail is required to reconstruct that in some digital form and how we would move through it. If you start working at a truly human scale, it can limit the size of the area that you can look at just because of access to resources and in your knowledge of that space.

And so there's a constant negotiation there, the kind of environment you build and the trades against what you can reasonably try to accomplish with your access to the right kind of computing resources and your ability to bend them to your will. And one of the things I discovered very early on in my graduate career was those push button tools were limiting. You know, they're 80% solutions for a lot of things. You know, you just want a quick answer. You can push a bunch of buttons, throw some data in, you can get an answer out, but they're effectively black boxes. Thank goodness that there's a lot more open source technology out there than there used to be.

You know, there's tools that archeologists can grab. You can go look at the source code. It starts to get daunting pretty quickly if you want to go beyond just trying to understand what these tools are doing and start building your own, but to explore things at the scale that we're talking about and to have as much fine-grain control over the questions we ask and the answers that we were looking for, you really do need to start building your own stuff. And that gets intimidating very, very quickly, but also do it in a way that's efficient so that you're not waiting years and years for an answer is hard. And, and that's where a lot of existing studies break down.

And none of us, all the researchers who are valiantly trying to do this, it's just, they run into these limits. There are limits in terms of training, exposure, proficiency in software programming, access to compute resources that it goes on and on. And so when we did this study about almost 10 years ago on Oaxaca, we sort of set a bar as myself and an archeologist named Sarah Barber who's at the University of Central Florida. She and I went to grad school together. We set this bar almost 10 years ago on what it would take to do what Stefani was describing in terms of modeling large-scale movement with fine spatial detail.

And the community looked at that and said, oh, that's a lot. We really can't do that. So we set this really high bar. And then what Stefani I did with this next paper was raised it again. To her point if you kind of tell him to the back pages of our article, she did all this work to kind of tally up on a per study basis. How many routes these studies have calculated? It's a rough metric, but she said, you know, there's tens, hundreds, thousands. You know, some of the more ambitious ones kind of get up into the tens of thousands. Our original study at Oaxaca generated over 4 million paths.

One we did several years later in the Mayan region was like 33 million paths, something like that. And then this was 125 billion if I remember correctly off the top of my head. 

Michael Garfield (20m 2s): It's daunting.To put that in context. One of the things that I appreciated in reading your research history in this piece was, I mean, most people are, I think, familiar with genomic models of migration and you make a really good point that the genomic models are limited by the resolution of genomic data, which is generational. So the timescale that you have is course, right? Like you're talking what like 15 to 20 years is your pixel.

You've expended that so vastly and in both directions. And you make the point that you're using human scale agents in this, but you're also looking at this continental scale, very granular model. And I'd like to hear you talk little bit about the features that you identified as salient features and the considerations that you applied, the design of this hypothetical agents, moving through the landscape here to give bookends to the very big and the very small that are all feeding into this 125 billion simulations that you ran here.

Stefani Crabtree (21m 13s): I think we need to jump forward to why did Devin and I start deciding to model within Australia and Sahul in 2017 in American summer. So Australian winter, I spent time in the Western desert with the Martu Aboriginal people, living with them, hunting, goanna lizards, doing all kinds of things. And I drove myself out from the airport in Newman to Barnwell and then had a rental.

And I was given the advice, if you get a flat tire, anything on the dirt road, don't go anywhere, bring a lot of water. And I was given this apocryphal story of these bank robbers who stole a car in Newman and then drove out to the Western desert and didn't have enough water. And they pop their tires. They wandered out and didn't know where they were going and they died. And then they were eaten by goanna lizards. And so I was given this story as this like scare tactic of don't go anywhere if you pop your tires.

And so it was kind of a little intimidated driving out there, but as this beautiful landscape, and once I got to the Western desert, it was very clear that something that seems inhospitable to someone who hasn't been there is somebody's garden for someone who's grown up there. I mean, it's the same thing for New Mexico where it's this beautiful landscape if you've spent time there, but the first time someone from New York comes there and you take them to tent rocks, they're terrified. They're like, oh my God, what if I wander off this path?

I'm going to die instantly. But if you've spent time there, you know what signs to look for. And so my experience with Martu, people of wandering within the Western desert to go hunting for goanna was starting to realize that they were queuing on to environmental cues. That were things that may not be easy to recognize when you're not from there, but that there was nothing in the desert that is surprising to anyone.Tthat was kind of an aha moment of trying to look at migration.

And my postdoc from 2017 to 2019 at Penn State working with Doug and Rebecca Bird was working in Australia. And so Devin and I were interested in looking at migration within this area. And we connected to Sean Ulm and Michael Bird, no relation, and Corey Bradshaw and Fred Trey and others who were interested in this question of the first people. So moving from my project for my postdoc was working with ethnographic data, people within last 50 years, but we then decided to start working 70,000 years ago. So what kinds of things can we understand about people and how people move and how people reason within their landscape, people today, whose brands are similar to modern humans who migrated out of Africa we'll say by 150,000 years ag., We have the same brains. We have the same way of reasoning.

What are the things that people would have come into contact with? And how would people have understood this greater landscape when they're encountering novel types of fauna. There is Australian megafauna at the time, but they still need water. They still need to move. They're moving on foot. So I interviewed some of the elders in the Western desert about how they moved within the Western desert until 1964, when they were moved to missions on the periphery of their areas.

So until 1964, they lived traditional nomadic life ways. And so I asked all these questions. 

How much water did you carry? I learned that they carried these bark cloths full of water, and it was about 10 kilos. I almost said 10 kilometers of water, 10 kilos of water. The most that they'd walked during the day would be about 30 kilometers. But you know, that's a little muscle may not be a depending upon the weather. And they would look to high points on the landscape to orient themselves.

Because a lot of the time you'd kind of be walking and you'd be hunting and then you'd go, oh, where am I? And you need to look up and you kind of orient yourself and you'd have this mental map. And this mental map would be kind of a top down thing. If you look at a lot of traditional paintings, they're actually landscapes. And they're showing you this dot painting of the world of how people would wander around within there. And it's based on what they see on the ground, but they have this piece together, this larger landscape.

And so we came up with these kinds of ideas of how they would walk and then who would walk. Well, it turns out that women, a lot of the times keep the family together. And so if we're going to model who is leading the pack, because they're going to have different caloric requirements, that's why we chose a 25 year old woman. She's kind of midway through her reproductive years, probably has some small kids with her that she's having to coax along. If you've ever walked anywhere with a four-year-old, you know, you need to coach them along.

If you have a child, you might be carrying it in addition to the water that you're carrying. So all of these requirements we then have to put into a digital person. So how does a digital person make sense of this novel landscape and how do they move throughout it and how do they choose when they're moving from one space to the next. And so Devin, as he talked about before developed this model called from everywhere to everywhere, which is literally that. So we're not looking at that one point to the one point. We're looking at on a pixel by pixel basis.

What are the costs of traveling from here to here and here and here and here, you know, looking at all of the spaces around you. So thinking about it like a checkerboard or something where we're not just moving like a rook, we're moving everywhere like the queen. That kind of formed the basis of how we were looking at these from everywhere to everywhere models. And then the other thing is when I say cost, I'm not meaning money, I'm meaning calories. What does it cost for you to walk uphill versus downhill?

Those are different, but they both cost something. You know, when you're going downhill, a steep slope, you might be breaking a lot. If you're carrying a lot of water that might make something a little different. If you're going across a Rocky terrain, that might be different. And so all of these kinds of things need to go into this model. And so that was kind of my fact-finding mission was getting all that data together. And then that's when we kind of came together and we're like, okay, we have all these ideas, now how do we put it together into this model, the largest ever migration model into a new continent where there are not digital elevation models. So we had to create those, how do we then model people coming in here?

Devin White (28m 22s): I'm glad you mentioned costs because you know, that's something I've thought about for years. It was one of the most humbling aspects of the early days of this research. And I started out as most archeologists do. You model the cost that you can understand? And it's easy to do this in terms of like distance or time, but that's what we're a lot of the tools do is what our GPS driven routing applications on our phones allow us to do. 

Stefani Crabtree (28m 45s): Andmore efficient route by two minutes. Would you like to take it?

Devin White (28m 50s): But there's traffic in that direction now. So it makes sense that that's where you would start. What I quickly realized though, was cost was going to be more complex and we needed to be thinking about more sophisticated conceptions of what cost means to somebody traveling, especially through a really inhospitable landscape and the caloric work that Stefani's referring to, it's interesting. It goes back to the sixties and seventies, interestingly enough, it has its origins in the us army. They were actually trying to understand caloric expenditure of soldiers, moving over different kinds of terrain, different environmental conditions.

So they put all these port kids on treadmills with different weights of stuff in their backpacks. And they tried to do as big a demographic slice as they could get in terms of just trying to understand the variability. So there are some limits that we recognize up front in terms of our ability to estimate caloric expenditure for non-Western non-military populations. But there are some aspects of the equations these guys generated going back again into the sixties and seventies that give you some reasonable insights as to what it's going to likely cost you if you're going to be walking at a certain speed on a certain slope uphill or downhill, carrying a certain amount of weight, and you're a certain age carrying a certain amount with you and over certain kinds of terrain.

So, you know, walking through sand is going to cost you differently than walking through a brush area in a forest environment, or just like on open pavement. And so there's different costs involved there. And, but even with that, I thought when I started this research, initially Stefani mentioned this earlier that I was working in this really inhospitable part of the Southwest. It's the Western Papa Berea, which is actually right on the border of Mexico. So it's, it's this part of Arizona along the diagonal. That is a really, really difficult place to be.

These days is actually occupied by the air force and the Marine Corps. It’s the Barry Goldwater air force gunnery and bombing range. A really interesting place to do your dissertation research, by the way. There's a whole set of stories there about just trying to navigate the complexities of working with any military bureaucracy to get out to this beautiful, pristine landscape and understand these ancient trail systems that are out there. And I thought, great. You know, I got this caloric approach. It'll make sense. People are efficient when they travel the economist fall into these traps all the time as well in terms of people being efficient, rational actors.

And I was able to model fairly successfully big chunks of this ancient trail system that I can actually see from space using satellites. But I would identify ancient trails. I would go out on the landscape, ground truth them, map them out, and then generate theoretical models based on these cost estimate equations. And to see if they would line up. It was sort of the early days of the research and there's certain areas that work beautifully, but there are these couple of areas where I was working with the archeologist on the range at the time, but Adrienne Rankin, who just knows that place inside and out, she's been there a very long time.

And I kept pointing to this one area. This area doesn't make any sense. Like the trails just aren't doing anything like I would remotely expect that people are trying to move efficiently. And she said, well, two things, one, you don't understand the landscape, as well as the people who've been living there for a very long time. And granted that's a very good point. And she says, also, you maybe need to think a little differently about what they are valuing in terms of what are they trying to minimize as costs? I spend enough time out in this region to understand that that area was what archeologists would call a ritual landscape.

That the goal moving around in that area was not about efficiently getting from point A to point B. It was about getting between these very important sacred places. And so it was a completely different kind of movement. So it got me thinking about how do you quantize that? How do you actually get that into a model where it's squishy or so to speak? It's not slope, it's not temperature. It's not time. It's the value people were putting on reaching a place and that place matters. And it doesn't matter how much it costs you physically to get there.

Sometimes making the journey more difficult is the point. If you think about a lot of pilgrimages that we even see today, like the whole point is for it to be difficult because it's supposed to be a struggle to get there. I always had this in the back of my mind since then. It was very humbling. And when stuff approached me about this project, and she told me about these interactions she had had with these Aboriginal communities and the things that they valued as they move it, brought this all back again. How can we try to encode difficult to encode things into a model at this scale? And so the visibility piece in particular was the most difficult that no one's ever really tried this before.

Because we knew prominent things on the landscape were important. We heard this from the people who were descended from the ones who first came into the continent and then it became, how do we actually like put that into a model? To some extent it was a geographic information-science to the rescue again, where there's a computational approach called Bewsha modeling where you can simulate a person standing at a spot, and then you ask like, what can they see? So anyone who's played like a first person shooter video game has experienced this where you're in a 3d environment and you're presented with the thing that you can see based on your physical constraints.

What we ended up doing though is running those sorts of calculations at the continental scale at regular intervals and kind of building up what we'll call a heat map of the most visible places from wherever you happen to be standing. So what would be visually attractive or visually approachable from anywhere? And so as you did that, you ended up kind of creating a probability model for the most attractive visual places on the entire continent. And then we figured out how to boil that down into an estimate you could put into the model so that a traveler as they were moving could factor that into their sort of internal mental map as, okay, I need to go this direction versus this other direction.

Because I can see that thing over there. And as a visual, a landmark in an otherwise very, very flat place. And it's something that's an anchor for me. That is probably one of the biggest things that we introduced into the model that took me the longest time to figure out what it was, how to encode that information, but it made all the difference in the world with respect to the results. We tried the more practical approach of modeling early on in this project and the results weren't lining up very well. But as soon as we went back and really thought about figuring out how to get the decision processes and the mental map of the Aboriginal communities, at least with this one factor of visibility, getting that built into the model.

So suddenly things started to line up much more nicely. And again, it's just very humbling to think through that. Anytime we make a travel decision, anytime we move, we're optimizing against a lot of things. Some of them are conscious, some are unconscious and how do we get a window into that? So we can build a better understanding of how people are moving. I'll get off my soap box here. 

Stefani Crabtree (35m 12s): I justwanted to add two little things to that. And one is that the heuristics that we're using for these people who lived 70,000 years ago, who are the ancestors of the people who I worked with, we're not saying that the people who are there today are a representation of these people. You know, we're not falling into that trope of anthropology from a hundred years ago, but we are saying is that there are similarities between how people reason and how people move about the landscape that are likely to be rather similar and things that we can use in a computer model to these people's great, great, great, great, great, great, great grandparents who came in there. And so that's where this beautiful view shed analysis comes in, where we know that everywhere people navigate, when you can't see where you're going, based on high terrain, if you don't have a map and you're put somewhere, you will orient yourself based on what you can see and people in the Western desert do that. And people probably did that when they first got into Sahul 70,000 years ago or so.

And I don't remember the other thing I was going to say.

Michael Garfield (36m 28s): That's just fine because actually between what you both have just said, there's like this very interesting kind of gem of correlations that I want to try and articulate and then pass back to you so that you can kind of riff on this. I love Devin, you bringing up this whole question about the tension between the remote sensing satellite piece and ground truthing something not just by walking the terrain, but by understanding how other people that inhabit that environment are walking that terrain.I'm reflecting on a conversation. I had recently with Andrea Wulf who wrote the Invention of Nature, the story about Alexander Von Humboldt and the origins of modern ecology and climate science. And she puts the narrative high point of his scientific adventure at the literal high point of his travels at the peak of Chimborazo standing at the top of this mountain, which by the way, indigenous guides led him there without which he could not have done this work.It's important to acknowledge that hisNaturegemälde,thismasterpiece of science communication, that he comes back with this showing all of the different vegetation that occupies different heights, different altitudes. There is this question about what it is that makes a site sacred. Why is it that it's the holy mountain and not the holy Arroyo? And it seems like it has to do with this relationship that we have with our environments and the use of visual cues. Like when JFK says we choose to go to the moon, not because it is easy, but because it is hard or George Mallory says he climbed Everest because it was there.

It is so prominent, so information bearing getting up there. The fact that most of the explanatory power seems in your models to have come from visual prominence is really interesting just as a sort of meta on how it is that you're doing the research for this paper and the way that you're combining. I was just reading some of the work by Lauren Klein of Emory University who came and gave a talk here at SFI earlier this year on data feminism, talking about distant versus close readings and like how you combine the distant reading of a text and the close reading of a text.

And there is a really interesting formulation. Living here in Santa Fe, the El Camino Ray, all this ancient trade route that now a huge stretch of that thousands of years old trail is Agua Fria. You know, it's the street that runs through the center of the town. And so you get these questions about the way that we paraphrase Winston Churchill, like we make the paths and then the paths make us. Our sacred maps of the landscape not just imposed arbitrarily by our emergent, from this relationship that we have with the landscape and with non-human entities, with the ecological engineering animals that are helping to shape that landscape. Thinking about the heuristics that you're talking about the visual cues, the fresh water least cost stuff, what this says about a general law for search. That's not simply true of humans, but it's true of agents in general.

And you know, how you navigate an information bearing landscape. And it lands in the kinds of questions that SFI economists like Sam Bowles, Wendy Carlin, and Suresh Naidu, and I do are gesturing toward with their core economics project where they're saying people only look crazy because you're looking at human behavior through this extremely chorus model that assumes basically like caloric expenditure rather than a much broader and multifarious definition of value that is actually this multidimensional value is the way that we encode much more profound information about the environments that we're trying to navigate.

Devin White (40m 28s): It's interesting you highlight that because, if you think about it, when you're traveling, it is a type of transaction. You are exchanging one kind of set of resources for some kind of angle. In this case, it's you moving to a different location. So you're constantly doing this optimization in your head. If you, even, if you don't realize that about, it's worth it for me to go from this place to this other place. And so there's a lot of parallels there too. I think economic theory about how people make decisions regarding things like near term versus like term goals and gains like where I'm traveling tomorrow could be different than where I'm going a month from now.

And as you're moving into a new landscape, that's a whole interesting set of factors to consider because there's just so many unknowns. But I also wanted to touch on something else. You mentioned a few minutes ago about one of the structural weaknesses of a lot of least cost analysis approaches is if you do these point-to-point analysis where it's just a couple of 10, 20, 30, hundreds of haves, you're doing these point to point transactions. You're getting these little glimpses into how people are negotiating with their landscapes and how the landscapes are affecting their movement and vice versa. And you get these tiny little glimpses.

You don't start to see the big sort of emergent pattern of that landscape and how it could structure human movement until you do something like what we did in this study. And that's really the origin of why we developed this from everywhere to everywhere approached so many years ago is we were interested not in any particular movement from one place to another, but from anywhere you might start, like, what would the landscape do to channel that movement? One of my former colleagues at Oak Ridge called it a pedestrian hydrology, like you're pouring a bunch of water on the landscape and you'd see where it flows. David Malakoff the writer for Science, he mentioned it as dropping a bunch of marbles on landscape and seeing where they go. That's really what we're after here is these general things that the landscape permits or restricts us from doing. And then how do we adapt to that? But then also how do we take advantage of what the landscape is offering to us and here's how we move around. And so it's like a negotiation and a dialogue as you go not to get too meta on you about it. The longer we've done this, the more we start to see this and you get away with those transactional studies and you really start thinking about these bigger problems, then you start wandering into other disciplines and then thinking about how these ways that we've explored optimized travel, depending on how you define costs, can impact studies in other things.

And I'm interested in stuff's thoughts on this as well. 

Stefani Crabtree (42m 45s): Whenyou brought up the idea of transactional and the things that you get of landscape, you're also giving something in return where you're creating a path and these paths have permanence. And we see that actually.

If you kind of look at the paths that we created, these superhighways the ones that were chosen the most often by people. When you first go through a landscape, you might be creating those footprints. Imagine being the first person snowshoeing through open meadow covered in new snow, and you can put those snow shoes anywhere, but then you come back at the end of the day and everyone's followed your path because you've made it easier for them.

Well, these people are doing the same thing. They're going across. Maybe they're following game trails of kangaroos are larger megafauna, but once those paths are there, they're going to keep being chosen. If they choose to not go on the path, that's an interesting thing, but people are going to choose that easier route. We call it the landscape architecture, the underlying geophysical constraints of the landscape make people move. And then people keep continuing those. You can see those paths and stock routes and the colonial routes from a hundred, 200 years ago, being kind of similar to where our super highways go.

We see this within our Sahul paper, but also you bring up Agua, Fria and the community and all that.  Whenever I run into people who are lost in Santa Fe, and they're having such a hard time. I'm like, you need to stop thinking like a car and start thinking like a Jesuit priest and a mule. You're at this Santa Fe and a lot of Northern New Mexico is very much constrained by a lot of the Jesuit movements and a lot of the indigenous movements and a lot of the pack animal movements.

And those movements are not going to be the same kind of straight roads like you get here in Utah that are optimized for white colonial settlers who show up 150 years ago or so. And so they're very straight, very rigid. You can count them, but in near Mexico, they're wandering because you're following the arroyo, you're following the high places. That style of route is very similar to what we see in south hall and are from everywhere to everywhere analysis. We make these 125 billion pathways.

These are how do people move throughout the entire continent over and over again. And I remembered my point from before. We're building up complexity, we're starting with the Knoll model, the knollest of knolls. And then we're looking at freshwater needs. We're looking at the needs for orienting yourself based on those high prominent places, which is how people would have navigated in Santa Fe 400 years ago, using the high places where you see Fort Marcy now. Avoiding the arroyos, especially that sand can break an animal's ankles walking near them, but not in them.

Those kinds of things are shaping the movement. And so while we start with 125 billion pathways, those aren't the paths that people are choosing. They're choosing some subset of that amount and those paths are being chosen over and over and over again. They're being inscribed and incised into the landscape so that we could still use them today. If I were to hand the map that Devin and I and our coauthors made to anyone and teleported you to the middle of Australia and said, okay, you're in Farmar.

I want you to walk to Perth. Here's a map. You could do it because we're using the geographic, the geologic underlying constraints of this landscape that have been there for 70,000 years and longer, because right, it's a continent that's been formed by billions of years of geologic processes. But if I were to give that to you, Michael Garfield, you would be able to take that and walk because you would know where the high points are. It might not be your favorite walk, but you'd be able to do it.

It's still there. 

Devin White (46m 46s): Yousee the same thing on the east coast as well. If you look at some of the earliest historic accounts for Western society settlers, those trails generally started out as a game trails that then the native communities followed and then inscribed. And then we started following them. Then they became wagon routes. Then they became roads. And now they're part of our infrastructure. And it gets interesting questions about how does that sort of transportation network structure the way cities look today. And it just keeps going and going. I was also reminded of this several years ago when we had the opportunity to take the speed approach and actually apply it to a modern-day situation.

We were looking at the movement of refugees out of Syria, absolutely horrific situation. And the question was, could we apply this technology in a way that you could create like an anticipatory map of where they might go based on their concerns and their worries and their fears and just what was open to them and try to meet them where they were even at borders and try to help them, so this is very much a humanitarian assistance and disaster response activity we were working on before I got to San Diego. And what was really interesting about this. It was several anthropologists working on this.

And one of them pointed out that some of these main routes that we were able to ground-truth, refugees really were taking these routes. And they've been important well into antiquity that many of these routes were ancient road systems through Syria. There were just persist to this day because they are just the most efficient ways to get around for a variety of reasons. And they've just stayed deeply inscribed in this landscape. And it was just the natural way for people to flow. So a lot of what we develop isn't just good for sort of projecting into the past, but we can also apply it to, to modern day and future scenarios. Not that I want to shift the conversation in that direction right now, but kind of just want to give you a teaser that we started to look into this and it speaks directly to what you would highlight it and stuffI brought up that many of these routes become deeply inscribed. And until you not only something about the human relationship to that place, but also you see it reflected in architecture and city organization and just how you move in general and even potentially how you think, because you're just in a different environment. You know, the Western U.S. versus the Eastern U.S. cities and built environment looks very different. Some of that is structured, I think because of some of these very ancient relationships between people in the landscape and how you move across it. And some of that's based on geology, some of that's based on other motivations for travel, as we've talked about.

Michael Garfield (48m 57s): This is reminiscent for me anyway of research awhole bunch of SFI people were involved in Mike Price, David Wolpert, Hajime Shimao, Brendan Tracey, Tim Kohler. It was a paper led by Jaeweon Shin. I hope I'm saying that right,scale and information processing thresholds in Holocene, social evolution, talking about the ways that we invent new information technologies. It seems from their analysis, it looks like major evolutionary transitions. We reach a scale of society that we need to organize new Infotech and that's, and that's where we are.

I mean, if you zoom out far enough, that's the story that we're telling about remote sensing and computation and a global civilization.From that point of view, it's just funny to reflect on how, I don't know who it was. I heard say this, but you can always tell what the most important feature of a society is by the tallest building.Used to be the church. The BBC's in our time has a great episode on this, about how arguably modern banking in the west started with the temple of Athena and then like borrowing your tributes back from the goddess Athena.

So the temple became like a loaning organization. And then now if you go to somewhere like Austin, Texas, the tallest building in the city is the bank. What we're actually seeing is we've crossed this threshold between the geographical features, having prominence to the society that the landscape agency sort of constrained and sculpted. Perhaps the least cost thing, since you did bring it up would be deeper applications of this model.

You did say let's not rush into the future here. So let's take a detour into the past and the controversy surrounding the peopling of the American continents. This is an area where I think a lot of people passionate about archeology have strongly held opinions about this. And it seems like your team might be able to shed some new light on this particularly contentious question.

Stefani Crabtree (51m 2s): There's somedebate in Australia and well people who study Australia on when people got there. People definitely got there by about 40,000 years ago, but the date has been pushed back quite a bit, but there still was it 50, was it 55? Was it 7 million? There's still debate there. And there will be debate there for a long time. There's a debate in the Americas about when people got to the Americas and how they got here, whether it was by land or by sea or both. There are people who are very staunchly,well, if it was by land, it couldn't have been before the Clovis technology because it was impassable.

But then there's this idea of people kind of voting along the shore. And there've been various archeological studies that have given us dates that were controversial. But about a week ago today being June 18th, there was a study that came out in nature that redated the collect Scotland cave study, where the dates that had some controversial old dates and the new study says 20,000 years old, which is very much so before the Clovis technologies showed up.

I think that hypothetical use of feet to look at people coming in by land and people coming in by sea, and then looking at the spreads of people throughout and looking at the demographic processes of the spreads of people throughoutand how quickly could you get people in these different areas like the Southern tip of Patagonia, where there are controversial dates of 20, 24,000 years old. I'm sure we'll be better dated soon. This would be a way to weigh in and say, people were likely here before, or they could have been if they followed these kinds of models. A lot of the time computer models are oldmodels.

And they look at ways of taking the fragmentary record that we have, whether that's of archeological data or looking at refugee migrations, things that are probable and possible that we don't have data for, or various other things like that. And kind of connect the pieces and say, if this is what happens, this is what you'll see. In a way, feet could be a way to ask those kinds of questions. If people were coming in by boat and we'd have to have a good model of that migration and then where those places of embarkation would be and how people would spread throughout and where that spread process would be.

And then we could apply some kind of agent-based model or cellular automaton model to look at people moving on those highways and how people would spread out. That would be one way to address the Clovis first controversy. I'm not going to weigh in on Clovis first. That's not where I work, but obviously like every archeologist, I have opinions based on my reading of the literature, but I think that computer models would be a good way to weigh in on controversy.

It's something that I've done before with the paper weighing in on controversy. There's been controversy about whether there was a hierarchical ritual structure in the Southwest, or if it was egalitarian and you can use computational models to look at that. And I wrote about that in 2017 and as a person who is a peace-loving person I like to always show that there's a middle ground. So I'm sure that the use of feet or something similar would be able to show that yes, even Clovis first and non-club is first people can get along.

Devin White (54m 40s): One of the things that Stefani mentioned about, the whole by land or by sea ideas is something that I've grappled with for more than10 years now, iswe built, pedestrian travel model. And if you're going to integrate any kind of maritime component to it, you have to kind of think about travel and movement and cost in a different way. And it's daunting, is something that even after we did the Oaxaca study almost 10 years ago, Stacey Barber said, well, what about modeling trade in canoes along the coast?

Because that that's a really important aspect of how societies at the time that were sort of in charge would have moved around. Like how do we do this? And I was like, oh shoot. Like, that's really, really hard.Modeling people just moving around on solid ground is pretty difficult. If you start putting them on a liquid medium with all these dynamic things going on, temperature and currents and everything, it just becomes very hard. One of the things that has come out of us publishing this paper, this has been some outreach in particular, from some research communities interested in looking at, for example, how the entire Pacific was populated. Getting to these remote places, getting to these islands, these little specks of land in the middle of this massive ocean landscape just requires a different way of sort of thinking through how people would move and peopling of the Americas kind of as a bit of a hybrid here because we would want to explore and would like to show if we went down this road in some fashion or enabled people to do sowhat these flows of people would look like coming through the land corridors when they're available, but also trying to, at least as much as we could accurately, I'm going to air quote here, accurately model maritime movement to this question keeps coming up again.

And again, is people move by water too. How do you incorporate that? Or at the point from a technological point of view, and also for my mental peace of mind that I think we could probably try to take this on. So I have this outreach after the paper came out from a group that was looking at how the entire Pacific got populated, which takes the Australia studywe did, and just blows it up even larger. I mean, it just makes my head hurt and thinking about trying to take on something that day,

Stefani Crabtree (56m 41s): People have looked at boat moving. I mean, there was a study at least cost analysis looking at how people could have gotten to Sahul and then a colleague of mine, M.S.Slayton who's at Carnegie Mellon for her dissertation looked at least cost analysis throughout the Caribbean, and has also looked at how could rice have gotten from mainland China to islands on a recent paper with Jesse Lasky, from Penn State. So people are looking at it, but it would be that amazing combination of the two that I think would really advance that kind of work.

And also just on the scale of the work that we did, the 125 billion routes, that is the kind of scale of analysis that would be needed for understanding the peopling of the South Pacific, since it came in different waves with different cultures, and there are now so many different language groups.

Devin White (57m 33s): It feels like the next frontiers is these sorts of massive movement questions in other parts of the world. I mean, we make that argument in the paper that we would really love to see the fundamentals of our approach picked up and applied in different areas. Part of it is to see how well they do, but also is just how generalizable are some of the things that we foundtoother cases of large scale human movement as a function of time across the globe, be it on land it in water or some combination thereof and keeping that big scale aspect to it rather than zooming into a small area.

Although it does work at that level too. You have to kind of tune this for the question you're asking. And another one of the modern examples that'd be, can always explore depending on time and interest, is it, you have something like a really detailed 3d model of a city. For example, one of the things we do in the remote sensing world is that we collect laser scanning data over large geographic areas, but it's a very, very fine resolution. And you can reconstruct three-dimensional landscapes, fine level of detail to the point where you could have a city in 3d, how would you model movement using the technologies that we've developed through that kind of landscape?

So, the computational complexities there is just you've narrowed down your geographic emphasis, but you can still explore these really fascinating questions. So the one case we did this that I've been involved in, it was after the earthquake in Haiti, when their infrastructure just blew up. Bridges fell, roads collapsed, just became impassable. And so everyone was moving around on foot through the city environment. And we need to understand, again, from a humanitarian assistance point of view, having the Google maps at that point, wasn't going to help us because that's not the reality of the people on the ground. And luckily that was at the Rochester Institute of Technology through a UN fund was able to get down there very quickly.

And they flew a laser scanner over Porter Prince. And Haiti came back with this really rich 3d data set, which is publicly available. Anyone can download it. And with it today, the whole goal was to do this as quickly as possible and make the data freely available. So we pulled this massive 3d data set in and ran a pedestrian flow model over the city in this new broken environment. How could people move around? Where might they be versus what the reality would have been prior to the earthquake? It was all about scaling your problem.

You can still apply these techniques even down to the scale of a city or a part of a city. And in some archeologists have done this for ancient examples as well. I'm thinking like Scott Branting, who's at UCF who has done this in the middle east. He was one of the early pioneers of the approach. It's exciting stuff. 

Michael Garfield (59m 60s): Just as an aside, there seems like an obvious link to work that Geoffrey West and Marten Scheffer did with MIT recently on the visitation law, finding those power law relationships in the places that people are choosing, be interesting to dovetail those, but I know that we have a hard stop. I really want to land this conversation in the upcoming SFI press volume, Stefani, that you've been working on,a textbook for agent based modeling for archeology and an article that you led in advances in archeological practice on outreach and archeology with agent-based modeling.

There's something that you said in this piece about the use of agent based modeling for communications with stakeholders. I know that in archeology, typically you're talking about cultural resource management. Given the obvious link between this particular paper and approach to studying migration, the conversation I had with Tim Kohler and Marten Scheffer way back in episode 33 on the shifting future human climate niche and their prediction of like billions of displaced people by the end of the century, that makes everybody a stakeholder.

So it really invites a set of kind of challenging questions about the way that this research is performed,the tools available to us to communicate it, when are the tools that we can use for research, the same tools that we can use to broaden popular literacy and invite stakeholders into a more interactive rather than merely receptive kind of relationship to this kind of thinking. Just as a nod to the indigenous piece of it.

I remember in Interplanetary Fest a couple of years ago, this conversation around, well, you know, for the indigenous dystopian, apocalyptic science fiction is regarded in the past rather than the future that they're already living in the wake of apocalypse. We're really into the big view here, which is when you start to see the systems completely enough, everyone is downstream. Everyone is a stakeholder. Everyone deserves to be involved. I'm curious if we can just close with your thoughts on where this hits the street and how you see the opportunity is for science and public involvement and this kind of thinking in this kind of research changing over years to come

Stefani Crabtree (1h 2m 21s): Well, archeology is a really interesting thing because in a lot of ways, everything that we are experiencing today and will experience in the future, somebody in the past experienced,whether that is mass migration of people or massive pandemics or things like that. Those are all things that we can look at data from the past to understand the present and the future. And I think that's one of the strengths of this paper is that we're looking at the possibilities of human movement across a large, somewhat inhospitable, unknown landscape, with different kinds of fauna and flora that people would have never experienced before.

And people moved on their feet. And this is going to be the reality of people fleeing, drowning coastlines in the future is that people are going to have to use their pedestrian movement to move from a lot of areas. Wealthy people will probably be able to take planes or cars, but if anyone saw the 1995 movie Independence Day, you know that the LA freeways get blocked up when the aliens invade. Well, it'll probably be relatively the same. If we have massive amounts of coastline change very rapidly. The ability for people to use their big brains to move through inhospitable terrain is something that we'll have to do in the future.

And so this kind of approach is very useful. Archeology is also something studying the past, studying our ancestors is something that most children have had an interest in. Whenever I tell people I'm an archeologist on the airplane, they always say I wanted to do that when I was a kid. And so you're right, Michael, the stakeholders are everyone, but computational modeling is a thing of taking what we know about the past and using these hypotheticals. And it's very game-like and it's a way for people to integrate their understanding of the past and watch it unfold.

And so in that way, the textbook that I have, and that paper that you mentioned are ways of trying to get more people involved and understanding computational approaches for social science applications. Archeology is not just digging in the dirt and finding a pot. It is understanding everything about humanity in the past. Archeologists have to be psychologists. We have to be economists. We have to be able to read these old army reports to understand how people move.

So in this way, the textbook is aimed at being veryaccessible for teaching people computational approaches and ways to integrate other kinds of computational work and social science work together to understand humanity. And so really, I do think archeology can save the world because we can look at the ways that people in the past have dealt with this in an ethical way, use these computational models to try and better our present and our future. 

Michael Garfield (1h 5m 14s): That was fabulous. It was really wonderful to talk to you both. I am left with so many more questions for you, but perhaps we'll have another chance to have you on the show. Thanks depo, Devin. And I could talk for hours. Thank you very much, Michael. Yeah, absolutely. Thank you.

Speaker 1 (1h 5m 36s): Thank you for listening. Complexities 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 communication efforts. Visit