 So, in this session, we're happy to introduce a terrific session called Studying Spatial Games in Open Simulator, and our speaker today is Professor Kevin McCabe, Kevin is Professor of Economics, Law and Neuroscience at George Mason University and Director of the Center for the Study of Neuroeconomics at George Mason University. He has numerous publications in experimental and neuroeconomics from his research over the last 40 years. His current interest is in the study of economic systems as computational systems, and he's been using virtual worlds in his research since 2005. So, welcome all, and let's begin the session, Kevin, over to you. Hello, everybody, I'm happy to be here. Excited to tell you a little bit about what we're doing in virtual worlds. As Scott mentioned, we started exploring virtual worlds in 2005, but it took us a number of years to get to the point where we could actually use it in our research. And so, we've gone to the point now where basically we're using it for everything, so we use it for research, we use it for teaching, here are a couple of things that we're doing. On the research side, and what I want to talk about today is spatial games, and I want to talk about two kinds of game designs that we've been considering. So, one design are what we call immersive games, where actions are meaningful, and in the second kind of game, we're interested in our more abstract games, where actions are symbolic. What kind of game design we use depends on the question we're asking. So, when we do immersive designs, we're very interested in the social conversation and the narrative that people form to make actions. When we look at the abstract designs, we're more interested in the strategies people are using to accomplish the tasks that they're trying to accomplish in world. So, this dichotomy, I hope to kind of describe with two games today. The first game is a more immersive game. And I'll talk you through a little bit of this game over the next few minutes. The important thing about this game is it's meant to simulate a commons, where people have to interact in order to solve a cooperative problem. The commons problem itself has been studied extensively by Eleanor Ostrom. And so, one of the problems we often encounter is what is called free writing behavior, where individuals do not do their fair share of work. This problem leads to all kinds of social dilemmas. And so, what we thought we would do is design an experiment to study this problem. When we implemented the experiment, we wanted it to be immersive. So, our goal was to build an island that had immersive features. And the island looks something like this. The point of the island, or what people are told who are immersed in this experiment, is that they live on an island with houses. The island is subject to frequent storms or hurricanes. So, we actually call it Hurricane Island. The storms come from the north or they come from the east. And what happens is the storms destroy the productivity of people who make money by having houses that are in good shape, essentially. And as the houses get damaged, they earn less money for people. So, that's their incentive. If you notice on the far right, there's a weather station. And the weather stations are used to protect the islands from the hurricanes. But there is a dilemma. If you're spending time protecting the island, you're not spending time earning money. And so, people have to jump in for the common good. And then there's also a coordination problem. Who should be defending the island? Not everybody has to. So, the question becomes, do they take turns or what? Now, it turns out this kind of immersive game can be studied using game theory. And so, what does the game theory look like? Well, it's a spatial game and it looks something like the following game. We can divide the houses into the... Here I'm looking at northern storms. We can divide the houses into the four houses at the top and the four houses at the bottom. Notice there's two weather defense stations, one at N1 and one at N3. The problem with this from a game theory perspective is that the people in the bottom third could defend themselves by going to N3, which is a shorter trip and less costly to them. If they do that, though, they're not defending the houses up above N3. So, as a consequence, this sort of forces people strategically in the top of the island to defend at N1, allowing the people at the bottom of the island to enjoy that defense without having to defend their houses at N3. So we call this a threat point or a sub-game strategy that people can use. I'm not going to go into a lot of details, but it leads to interesting dynamics when you actually run a game like this. And I just want to spend a minute showing you a chat text where we can see this dynamic and play. And so I'm going to show you two slides with chats. This first slide is interesting at showing you the conversation where the group has decided on a strategy, but then they started criticizing the strategy. They didn't like it. So they were having problems. And in the process of trying to figure out what to do, you can see that Green Gardener here is actually talking about the sub-game strategy to go to N3. A little later on, this leads to the homes getting damaged. This group was never able to actually solve the coordination problem that was required because of the early lack of accepted planning. So anyway, as you've studied these games, we've done this now about 20, 30 times with subjects that are being paid actual money to participate in the experiment. And what we find is that basically lots of groups can coordinate, but they have to do it by coming up with an early planning strategy. And what we find is the planning strategy should, in fact, involve the Nash equilibrium of the game. That is, they should be planning around the strategic incentives of the players. And I want to now switch to an abstract game. This is a spatial game we've been developing. And what I like about this abstract game is it's a platform we've been developing that allows us to design, or allows anybody to design games without knowing a lot of programming. The immersive games need more programming. And in general, spatial games have become very interesting since Thomas Schelling introduced them again. I'll tell you a little bit about the package. I want to make it available at the conference to people. It's really a beta version, but most of what I'm going to talk about today is working. So the first thing you have to do in the package is you have to design the game that people are going to play. And the game is built on a HUD. So basically, you design the payoffs in the HUD, what payoffs you want. And they're put into a payoff matrix. And you can do that directly on the HUD. The payoff matrix here, the example we're seeing, there's two players. One player has a choice of X or Y. The other player has a choice of A or B. So this is a standard normal form game and game theory. Once they make their choices, they're going to choose an outcome. The outcome will determine the payoffs of each of the players. The bottom left payoff is the row player and the top right payoff is the column player. So this is the kind of game you can design in the setting. Once you have several games designed, then what we do is we have a system of periods where all the players are put in boxes above the island, separate boxes. And then at the beginning of the period, they drop down onto the island. And they have to find each other and match with each other. So in the matching process, they're looking for a player to play with. And initially, there's lots of players they can play with. And the numbers on the right tell them who are nearby that they could choose to play with. And they can send a message by clicking on one of those numbers to tell the other player they're ready to play. Now when they finally choose somebody, which is this gold button lit up, then they can play the game. And there's several games that could be played depending on how they've been matched with each other. And so let me jump to the next slide. And I noticed somebody mentioned Kay's work on Prisoner's Dilemma, and this next slide will kind of show you an example of that. So here's an example of a game where if you're matched with somebody with the same color shirt, you play the game on the left, which is a coordination game. If you're matched with a different color shirt, you play a Prisoner's Dilemma game. The coordination games are easier to solve for people and therefore more comfortable. The Prisoner's Dilemma games are harder to solve and therefore less comfortable with for people. And so the question is, is this going to affect the way they choose to match with one another? I'm going to show you a moment of data from an example of high school students who played the game as a pilot. And what we see here, this says different shirt outcomes, but it should say same shirt outcomes. Sorry, that was a mistake. But what you see is out of 160 matchings that took place in this game, 64 of them were same shirt matchings. And 26 of those 64 pairings were able to coordinate on the high equilibrium of 10-10. We can compare that to the people trying to do the Prisoner's Dilemma, which is a higher payoff game for everyone, so they should be motivated to try. And what we find, if they match with different shirts, they try that 96 of 166 times. But you see a fair amount of defection. They're only cooperating 25 of those 96 times. What's even more interesting is if we break this down between first and second period pairings, what we can see is that basically in the first half of the experiment, only one of the pairings was able to cooperate if they went to the Prisoner's Dilemma game. So it takes time and effort to develop this cooperative strategy. Now over time, if we look at the second half of the experiment, what we see is greater cooperation and movement towards that. This was, by the way, in a playing field where there were no what Schelling called vocal points. Adding them could actually improve cooperation. But the nice thing about this game is you can set it up by just changing the payoffs, changing the matching to game rule, and then play it. So in fact, students with very little experience with programming can build a virtual world game within a couple of hours and be playing it with each other. With that said, I'm going to end now and see if there's any questions that people would like to ask. And thank you for your time. Great. Thank you very much, Kevin. So let's take the opportunity to see if anyone in the audience has any questions they'd like to ask. Okay. Well, maybe I'll pick off the questions myself. So, Kevin, I'm very interested. I think at one point you just mentioned there were some high school students involved. I mean, you were doing these experiments from the perspective of research, but in terms of the people that participated, what kind of students were they if they were students and how did they respond to the two different types of games and that approach that you adopted? Sure. What we find, we run a high school program once a year in the summer. At last one week, we usually try to bring in students that have a computer science background or interest. But often we get students that have very little computer programming background. So if they have a good programming background, we can get them to build an immersive game. And we usually try to have them do some kind of simple immersive game. I think the interest that comes from immersive gaming is, of course, the social aspects of the game. There's a lot of social learning that goes on in immersive gaming. But if you don't know how to program, there's another learning aspect to this, which is really good, which is learning how to build a game to study a problem, learning how to build a system that does something. And I think there it helps a lot to have alternative mechanisms like the Spatial Game that I was talking about at the end. We have a couple of other questions. Cyber Serenity Vela is asking, where can we look at the game or games? Yes, so I have an island on Kitely at the moment. We're doing a lot of research, but I'll go and it's under the center for the study of Neuroeconomics, so they're all front-ended with CSN. I'll go ahead and set up a few islands later today or tomorrow that has some of these examples on it. But I'm also very interested in getting people's feedback if they try to use the new Spatial Game or if they're interested in helping develop some of the software behind it. So what I'm going to do is leave a builder's box as soon as I can figure out how to get it in-world at my exhibit and people can take it and put it on their own island and it should be clickable and build and allow you to try to do it. Lisa, thanks for saying it perfect. There was another question from Delightful Do-Angle. What is next in the research plan? So we're interested in a few areas of research right now, but one of the ones that I really want to do with the Spatial Games is go back to something that Thomas Shelling talked about a lot and built a game for and see to what extent we can look at this happening in experiments. And this is this problem we're seeing in the world now of coalition or click formation. I think some people call it tribalism, but this idea that people tend to congregate into like-minded communities and don't take risk learning from and trading with other communities. And I think it's a really important question. And once you start doing experiments where you can get the phenomena to occur, then you can start to ask, well, how do you change it? How do you make it so that people are more willing to engage across communities? And that's, for me, one of my more interesting research questions. Thank you, Kevin. Juliet, surreal dreaming was asking, where is your booth located? Graham has responded just so that everyone sees it. Expo zone three. Right. Give me a day to get the box over there because I'm afraid that it's going to take me a moment to figure out how to import things. I just figured out I had an Expo booth yesterday. So I'm a little behind everybody, I suspect. I think I'm booth number seven. OK. Well, I think we'll have to wrap up here. Kevin, thank you so much for a genuinely fascinating presentation. And if everyone, let's offer our applause to Kevin. So thank you very much.