 All right, so the second part of the session will be concerned with simulations and network science. And that's why we decided to have a bit of an introduction so that everyone's on the same page and everyone's actually can have a good understanding of what people will be presenting. So we'll start with some basic definitions. And the super basic was is complexity science. Complexity science is a branch of science that is concerned with studying complex systems. And when you, as an archaeologist thinking about a complex system, you probably mean complex hierarchical societies. This is not the definition that is used across the science. So let's look what a complex system is using their words. So a complex system is usually simple. It doesn't have to be complicated. But even though it's simple, it's very often very difficult to predict. Most of them consist of many elements, and those elements interact with each other. And their interactions still may be simple, yet the results of those interactions are not. So when you think about it, the most obvious example of a complex system is a human society. It consists of many elements, humans, who interact with each other. And those interactions may be pretty simple. There may be communication of ideas. They may be passing of information from one person to another, or, I don't know, getting together and forming a household. So those interactions not necessarily have to be complex, but the results of them, the pattern that we have in economy and social changes, they are very difficult to predict, and they are not obvious from just by looking at the characteristics of the people, of the entities that actually make the system. So that's a complicated subject, so let's give an example, right? So here are neighborhoods, Jewish Arab presidential patterns in two neighborhoods in Israel. And when you look at them, I don't know who's whom. Let's assume the Jews are green and Arabs are blue. When you look at this, the logical conclusion is they absolutely hate each other. And this may be the case. We don't know, we don't know. But maybe there's another reason why they are so segregated, why they don't live together. So quite a few decades ago, Schelling, who got a Nobel Prize for that, created a model, and the model is very simple. You have triangles and squares. You can call them Jews and Arabs, but you might get into the foggy area. And they only have one pattern of behavior. They basically have preference on how many of their neighbors should be the same as them. They can be like, okay, I want one in six of my neighbors, like I want my neighborhood to be at least one third the same as me. Or you can be like, oh, I really want that it's more than half. So if you then allow them to move every time they're unhappy, if they can change the position of their household, then you get some really interesting patterns. If their preference is for 10% of people living around them, then they're pretty well mixed. But you can see that already, when they have only 35% preference for people like them to live around, you end up with segregated neighborhoods. I mean, to say that like three people in 10 of my neighbors should be similar to me so that I'm not a crazy minority, to say that those people hate each other. I mean, that's quite an overstatement, right? And it turns out, you know, if you basically just don't want to be a minority, you have a super segregated neighborhood. And the only way of not having segregated neighborhoods is to basically wanting almost everyone of your neighbors to be like you, so being pretty racist, right? Because then nobody's happy and everyone just moves around. So you can see that it's funny that like, even nice people that have no problems with the other type of people are handed up in this social situation that we interpret in a pretty dramatic way. Without using complex system approaches, we would have never guessed that. At least that's what the Nobel Prize Committee believes. So the most common tool for studying the behavior of complex system is simulation. And the most common way of representing the structure of complex system is network. And now Tom will give you the pitch about networks and he's got exactly, and yeah, and the two, they make complexity science. He's got exactly three minutes for that. Three minutes, thank you. I can move your slides. No, no, no, thank you, I'll do that. So yeah, network science is definitely one of those main approaches that we use to study complex systems. Network science allows us to deal with network data. And I very often like to think about how we use GIS as archeologists as a comparison. We use GIS to deal with spatial data. We make maps, we represent spatial data, we manage spatial data in our spatial databases, and we analyze spatial data using GIS systems. And they can do all of those things. Now, network science does exactly the same, but with different type of data. It does that with network data. We use network science to represent network data, to represent dots and lines. We use network science to manage them in our softwares and our databases. And we use network science to analyze that network data. Like in GIS, we ask questions about the relationships between places, between sites, not based on their spatial proximity, but based on how they are related to each other. So we could ask questions about how central is this particular site in this network? How important is this place as an intermediary for the flow of information, given the relational structure between the different sites? These techniques are most commonly applied in archeology to represent our archeological data as networks and to explore the implications of that, or to explore the structure of our network. But also to represent our theories about how we feel or how we think past societies might have interacted with one another as a network model. So these are two examples of those. Now, there's a couple of fields of application that have emerged as proving that network science can make useful contributions to archeology. They tend to focus on road networks on the one hand, which mainly concern those cases from, for example, the Roman Empire or Inca societies where you have physical evidence of actual roads or very well documented information about roads, and we can explore how information and goods and people and the military flow through those, but also network representations of least cost-pass studies that are performing GIS. Another prominent field of applications is that of island connectivity. So for example, John Terrell, back in 1977, theorized that Pacific Islanders had a limited ability to interact with one another. They couldn't interact with all of the islands around them, but rather tend to interact preferentially with the three nearest island communities. So he represented the network model to represent that theory. Also in urban studies and architecture, where we use the techniques like access analysis to represent urban spaces or house spaces and the connections between them, the ability to move from one to the other as a relationship, which is basically graph theory and network science. We also deal with social networks. Now here, this is a perfect example of like archeologists use either data for networks or theories. How many cases do you know in archeology in which you have data about social networks? I happen to know a few because I do Romans and we are just spoiled with data. It's a fantastic place to work, but like the vast majority of archeologists don't have data like that. So very often we have a theory about people in the past that interacted with one another, that created communities, information was spread throughout those communities and that we theorize implications of that. So we use network models to represent our theories. Now, another big body of applications are similarity networks, representations of archeological site assemblages and their similarities. And this kind of application where we represent sites and the poultry for example, the artifacts that are part of the assemblage, we draw a relationship between an artifact type if it's part of that assemblage. This to me is a good representation of the diversity of things we can do with network science. We can just use it to represent very typical archeological data and explore our data set in a different way as a different data representation. But we can also formulate theories about why these relationships matter. We could say as we so often do in archeology that if a pair of sites has some artifact similarity then in the future, those sites will have more increasingly similar assemblages. So we assume that a relationship from site two to type one is probable to happen in the future. This is just an archeological theory about what these relationships mean and why they matter. But these are things that we can express and evaluate with network science. Back to me. That was perfect, three minutes. No, two. So we go back. So that was how we usually describe the structure of complex system. We use network to do that. In order to understand the behavior and the outcomes of interactions in a complex system, we mostly use simulation. So to go from the super low level, let's start with a model. So a model is a simplified version of a real system of something that actually exists in reality. So you have a fire truck here. And this is a model. This is a drawing of this fire truck. It's not a fire truck. Fire trucks are not one two dimensional and they're not made out of paper. This is a model. My kids has a similar one. This is a model of a fire truck and it's made out of plastic and it is red, right? So it has some elements of the original system that are similar and other elements that are different. This is a model. Again, it's red and it's probably more realistic than a plastic one, but it also doesn't exist in the real world because it's a game. And this is also a model of a fire truck. It's probably a very exact simulation of the use of petrol in a fire truck. So you can see that every system can be represented by many different models depending what you're actually interested in. If you wanna study the fuel consumption in a fire truck, you're not gonna take my kid's plastic fire truck toy. But if all you're interested is that it roughly looks like a real fire truck and it makes noises that are really annoying and it's red, then that toy is exactly what you want. If you want to build a model and understand how it changes over time, then what you're gonna end up is a simulation. Simulation is basically a model with a time arrow that goes through it. So we use it to model processes. We wouldn't use it to model a fire truck, but we would definitely use it to model how the fire truck behaves on a road, for example. And when do we use it? We use simulation when we're entrusted in a process, but we cannot directly observe. We cannot directly observe it because it's too big or too small, like say, I don't know, changes in galaxies or super tiny things inside an atom. We use it if we cannot observe the process because it is yet to happen. So for example, simulations are used to model how to change your taxes for next year. You know, they don't do it just at random. They actually run a simulation first to check how it's gonna impact on the economy, at least we hope. And Ori, for example, I'm sure this building had a simulation done to see what happens if fire starts. Can people actually escape? So all those things are done because you cannot just do an experiment. And is there something that we would really like to know more about, but we have no access to it? The past, right? I mean, this is like the most obvious example of a system that happened and worked and we're super interested how it worked and we have no way and we'll never be able to understand, you know, just go and observe it. All right, so we have to speed up. Why model in archeology? There are three major reasons why you would like to do it for theory building. It's basically if you're developing your theory instead of writing an essay, you can use a formal method such as simulation to basically see whether what you're proposing will actually have, will produce a data pattern that you see. So for example, if you're interested, okay, let's see, I have my calendar that I study. What if population grow is high? Would I have such an urban sprawl as I see in the data or not? So you can use it to just basically check whether a theory is internally consistent and what would be the data prediction coming out of it? It's perfect for hypothesis testing. You represent, you know, like we have many, many questions where there's quite a few explanations that already sounds pretty plausible. And to choose between one and another, you basically choose on the basis of the color of the eyes of the person that proposed it. It's basically random. One of them is probably right or there's true somewhere in the middle, but there's no way of saying hypothesis testing can be done using simulation because you simulate every single of those hypotheses and then whichever one is the closest to the data, gives the output, the status, the most similar to the data, you give it the first place for now because then people get upset and they change their hypothesis and then you have to do it again. But in the process, you actually get the hypothesis more and more refined and closer to hopefully truth. And then middle range research. Is this idea, you know, like Binford went and looked at hunter-gatherers and then looked at the residuals of their camps and then compared to archeology with the idea that once you have access to a real system that you can observe and the results of their actions, you can compare those results to archeological material and you can conclude that probably if they're similar, the people in the past were engaging in similar behaviors. We'll look at it in a bit more detail here. So when we do data analysis, we have our pods and stone tools and things and we do a description of it. We use typologies, we use statistics, spatial analysis, all sorts of things in order to see the grand population level patterns, the distribution, the change per time, to basically trends in data. When you run a simulation, you build a model where you understand all the interactions and that model has to produce what we call artificial data. And whoever was in the first part of the session, you could see that even 3D models do it. They create this artificial data. In that interfacial data, we'll have the same structure as the real data. So we compare that to and you conclude that the behavior that you modeled plausibly could have produced this type of patterns. So this is the middle range research type that you can engage in using simulation. There are different types of simulations, which we don't use because we really shit in maps. We use agent-based modeling. And there will be a lot of presentations about it. Agent-based modeling uses tiny software units, agents, and they can be people, they can be sheep, they can be cars, they can be anything you want. The thing is that they're all heterogeneous. So every single one can be different. They can have different age, different gender, different energy value, different culture, different whatever you want. They interact with each other and they interact with their environment and they make autonomous decisions that are deterministic in a sense of rules of behavior, but they're still autonomous. So the purple agent may decide, I'm gonna stay here and sit on my patch of food and eat it, whereas the blue agent is probably just looking for some food. So every single one can behave in the way that is best for them at that particular time. And they learn and adapt if you're a good coder. So summary. So issue whatever is actually much more complex and previously believed is the most common conclusion of every archeological research I've ever seen. And this is definitely true. To which Andre Costopoulos says, simulation is our only salvation. There are no other methods that can give you causality of processes in a formal field. Everyone else uses it and it's time that we just get on with it and do the same. So if you're interested, there's some links or come and see me after the presentation. Thank you very much. Thank you.