 Hello, everyone. This is Alice Gao. For the next few videos, I'll be talking about game theory. Game theory is a branch of microeconomics. It is a topic that's near and dear to my heart, because it's closely related to my research. My research is at the intersection of computer science and microeconomics. It's a field called algorithmic game theory. The funny thing is that I did not take a single economics course as an undergraduate student, and somehow I got into research on algorithmic game theory. So sometimes you never know where life takes you. You just have to go along with it and enjoy the ride. Normally, I would start by asking this clicker question. Have you learned game theory and or mechanism design in another course? Obviously, I can't ask you this question right now. At least, I can ask you, but I can't get your answers. So if you have learned game theory and or mechanism design in another course, this will be largely a review for you. I will only have a very short amount of time, so I would only cover the most basic of game theory. I'll talk about two-person normal form games, and I'll go over a few solution concepts. Dominant strategy equilibrium, Nash equilibrium, Pareto optimal outcomes, and talking about how to derive mixed strategy Nash equilibria. And that's it. If you have learned these concepts, you can probably go through the videos quite quickly. Now, if you haven't, please enjoy. This is one of the topics I really like, so I hope you also like it. This course is about building intelligent agents for an environment. So far, we have assumed that we are the only intelligent agent in the environment. But what if this is not true? What if there are other intelligent agents in the environment as well? These other intelligent agents may have their own goals, preferences, and they are also reasoning about what to do. We should not treat these other intelligent agents as noise in the environment. When we're making our decisions, we should take into account that these other intelligent agents are out there, and then make our decisions based on our knowledge about how they would behave. Game theory is an important tool we can use to reason about how agents will behave when there are multiple intelligent agents in the same environment. If you ever try to explain to your friends and relatives that your research is in game theory, chances are they would start talking to you about games. Actual games like video games or board games, and that is if they enjoy playing games. Otherwise, they probably stare at you weirdly and think, do you play games all day for a living? Game theory is this weird term that most people misunderstand it as only about games literally, but it actually is much more broad than just about games. So on this slide, I just wanted to show you some nice pictures of games, but as a side note, Hanabi, this is one of my favorite board games. It's a cooperative game where all of you are cooperating to try to achieve a common goal, but you cannot see your own cards, so you need to communicate with your teammates to achieve the common goal, and I really like this game. Ghost Story is another cooperative game that I really like, and I played while I was in grad school. So you can see that I tend to prefer cooperative games than competitive games because I really dislike confrontation and negotiations, so cooperative games, I like them better. As I was saying, game theory applies much more broadly than literally to games that we play. In fact, game theory applies to most situations in life where we have to act strategically. A game here is not just literally one of the games that we play, a game here is a mathematical model of a strategic scenario. So for any scenario, as long as there are multiple intelligent agents interacting with each other, we can call it a game. In fact, you can think about life as a big game, and we're all just pieces in this big game. Let me give you a few examples of applications of game theory and mechanism design. The first example is the Dutch flower auction. You can see in this picture here, so every morning fresh flowers from all over the world get shipped to the Netherlands, and the flowers will be sold in one of these auctions as happening right here. You can see the clock on the walls. Those are displaying the current prices that the flowers are being sold, and then there are buyers out there who can look at these prices and decide on whether they want to buy these flowers at those prices. So once the flowers are sold this way, they again get shipped to their destinations. An auction here is simply a mechanism for resource allocation. Here, we're trying to allocate flowers through these auctions, but there are all sorts of auctions happening in the world in all sorts of shapes and forms. For example, you may or may not know that the advertisement displayed in Google search results are also in automatic auctions. An auction is a type of mechanism that are widely studied in algorithmic game theory. The second type of problems we can use game theory for are called matching problems. I've given some examples here, but let's talk about the medical residency matching problem. So once students graduate from medical schools, they need to go to a hospital to be a resident for a few years before they become an actual doctor. If you ever watched Grey's Anatomy, you should know this by heart. The entire show is all about new grads from medical schools going through their internships and then residency and so on and so forth. Now the problem here is that each student will rank a bunch of hospitals, and then each hospital will rank a bunch of students, and we want to match the students and the hospitals such as such that the resulting matching is stable. So what does it mean for matching to be stable? Well, the matching is stable if we cannot find a pair of student hospital such that both the student and the hospital are happier if they switch to a different choice. Okay, this is called the stable matching problems. There are some similar problems, similar matching problems. For example, school choice, that's the problem of matching students with secondary schools. And even organ transplant, that's the problem of matching kidney transplant donors with patients. So in all of these problems, we have multiple agents, the agents have their own preferences, and we're trying to match the agents in some way so that they are happy in some sense. The third type of problem that uses game theory is crowdsourcing. Crowdsourcing is based on the idea that the crowd knows all the useful information. All we need to do is find a way to extract the information from the crowd. I'm showing you a picture of the ESP game here. The ESP game is a relatively old example from a few years ago. For this game, the original purpose is that we want to label images, but at the time, it was difficult to do so automatically. So can we get people to label the images for us for free? Well, some researchers did design the game where two people are randomly matched, and the goal of these two players is to guess what the word the other person is going to type. So hence the name ESP, right, six cents, you're trying to connect with the other person in the six cents and guess what they're thinking. If they match on a word, then the two players win the game and they can move on to the next game. Now in this game, the two people cannot communicate with each other anyway, other than the fact that they can see the same image in front of them. So naturally, the two people can try to match by typing a word that's relevant to the image. This game turned out to be quite fun and people enjoy it. As a side effect, they generate labels for these images. Unfortunately, this game is no longer available, so you don't really have a chance to play. Now there are many other examples of crowdsourcing, and it is believed that as long as you can design a website in the right way, you can get the crowd to do anything for you. They can design things for you, like 99 designs, they can write code, like top coder, they can translate text for you. Duolingo is an example. And finally, you can even get them to rate courses and professors, like UW Flow. That's everything for this video. I hope the examples made the concept of game theory a little bit more concrete in your head. Thank you very much for watching. I will see you in the next video. Bye for now.