Nash equilibria – benefits and shortcomings
https://www.newyorker.com/news/john-cassidy/the-triumph-and-failure-of-john-nashs-game-theory
This article contemplates the successes and failures of Nash Equilibria in describing phenomena. Game theory had existed prior to John Nash’s research however it only explained outcomes in zero-sum-games such as poker in which one person’s gain is another person’s loss. Nash’s discovery built on this idea of game theory and allowed for situations to be explained that did not have a winner or loser. For example, a set of farmers are dividing up a market. This is not a zero sum game, it is rather dividing up profits, and Nash was able to explain these phenomena as scientists in the past could not accurately.
Nash was able to prove that given a finite number of players and a finite set of moves, there must exist an equilibrium. The article goes on to posit that Nash’s theories had serious drawbacks in usefulness for several reasons. One being that it is only applicable in a setting with a finite number of moves and a finite number of players. Such a situation is not always common in the real world where people will adhere to such strict rulesets and defined actions. Another drawback is that there could be several Nash Equilibria with equal value and no way to determine which equilibria will be selected. This is a drawback of a theory that was designed to pick out particular solutions.
The article suggests that we ought to use Nash Equilibria for what it is good at whilst acknowledging its drawbacks. Nash Equilibria are good at suggesting what will not happen, and we can make decisions from this information. This tactic is used often by legislators when proposing if a new rule would have a desired affect. Such applications of Nash Equilibria are helpful. The article suggests that we acknowledge the flaws of the theories pushed forward by John Nash whilst using his breakthroughs for what they are truly good at.
“The War Between Huawei and Apple: A Convoluted System of Networks”
Article Link:
https://www.cnn.com/2019/08/26/tech/huawei-futurewei-us-future/index.html
In the CNN article titled “Huawei was Poised to Fight Apple in the US. Now its Fate is Uncertain” the author, Clare Duffy, discusses Huawei’s attempts to utilize the United States global market in order to gain a larger technological following. Huawei, which is based in China, is determined to expand its smartphone commerce by competing with the Apple iPhone in the United States. According to Duffy, Huawei, “the world’s largest telecom provider and second-largest smartphone seller” is using Futurewei, Huawei’s “US research and development outpost” to invest in rising research and development organizations. This technique allows Huawei to utilize United States technological resources to gain insight on innovations that are gaining momentum in the field of tech. Unfortunately for Futurewei, the White House has made it difficult for the company to maintain American influence by banning the company from transporting tech machinery to other countries. The company is also barred from “buying parts from important US suppliers, such as [Google and Intel].” Due to this, Huawei now has limited staff, dwindling partnerships, closing branches, and weak ties to the world’s leading technological countries.
Meng Wanzhou, Huawei’s senior official, states that the United States and Canada have played a role in the company’s struggles. As a result, the company is currently considering whether or not it would be beneficial to continue investing in American research technologies. On the other hand, Huawei has lost partnerships with several American research universities, due to governmental restrictions. The goal of these partnerships is to create new technologies that have the potential to pave the way for future advancements in the world of tech; this would in turn, help with Huawei’s public exposure. However, as the business continues to lose funds and societal support, it is unclear if Huawei will be able to maintain its title as one of the world’s leading tech innovators.
Perhaps it would be helpful to view Huawei’s situation through the lens of societal network properties. To begin, the Strong Triadic Closure Property suggests that Node A (in this case, Huawei) violates the property if node A has strong ties to nodes B and C and there is no edge between nodes B and C. I have decided to label nodes B and C as Google and Intel, respectively. For this example, it is evident that Huawei has weak ties with both Google and Intel, as it is blocked from doing business with these companies. Google and Intel do work together for research and development, so we can consider the ties between Google and Intel to be strong. Because of this, the Strong Triadic Closure Property is satisfied.
Another one of Huawei’s societal networks can be interpreted by using the Structural Balance Property. According to this property, the edges between every set of 3 nodes has to have 1 or 3 positive edges; a negative edge would represent an enemy or an opposing force, whereas a positive edge would represent a friend or a friendly connection. In this network, node A would represent China (Huawei), while nodes B and C would represent the United States and Canada, respectively. According to the article, China (Huawei) is currently having issues with the US and Canada, so the edges BA and CA would have negative edges. The United States and Canada appear to be friendly, so edge BC would have a positive edge. Luckily, this network satisfies the Structural Balance Property.
Of course, Huawei’s struggles can be depicted through various network systems, but I have chosen to highlight the main networks mentioned throughout the article. During the past few lectures, I found the Strong Triadic Closure Property as well as the Structural Balance Property to be particularly interesting and soon realized that they would be extremely applicable to the subject matter discussed in the text. Furthermore, I would say that I was pretty unfamiliar with the influence of Huawei’s tech innovations prior to reading this article; I had only heard about Huawei in the summer when I visited Portugal and was bombarded with several Huawei advertisements. Overall, I would say that my analysis of this text has helped me further understand the impact of networks on the technological society.
Social VR and Our Virtual Social Graphs
Social VR could be the means in which we strengthen the otherwise weak connections between us and our virtual companions. Currently, when we ask someone to describe their “virtual network” they probably respond with the number of friends they have on Facebook or the number of followers they have on Instagram. Although number of friends / follower count accurately describe the magnitude of their virtual social reach, these metrics do nothing to describe the strength of our social connections. In fact, in many cases, these numbers do not even closely approximate answers to questions like “how many close virtual connections do you have”. Many of our online connections do not offer strong social value, but VR is aimed to change that – VR can effectively shrink our virtual social presence but strengthen our social connection.
In social VR experiences interaction with another user is not mindless scrolling through their photos, dropping a like on one of their photos, or leaving a cliche comment on their posts. Social interaction in VR are truly present, human interactions. In social VR you are speaking face-to-face with another virtual user, using fully animated body language, and socializing approximately as one would in the non-virtual world. This sense of social presence will reinforce our connections with those we decide to engage with, but the time and intentionality involved will make these types of interactions far rarer than current-day online social interactions; consequently making the size of what people describe as their “virtual social network” much smaller but much stronger.
Programs Playing Poker: Pluribus and Game Theory
https://science.sciencemag.org/content/365/6456/885
Game theory lends itself nicely to games such as rock-paper-scissors, simple attack-defense games, and certain problem set questions from Cornell University. Algorithms already exist that can guarantee convergence on Nash equilibrium for any two-player zero-sum game (CFR being the one that will be touched upon later), so one can easily see how applicable this facet of game theory can be in practical application. However, the significance and strategy involving hidden information in poker has made it an immensely difficult game for artificial intelligence to model. While poker has forms of many differing complexities, this article details Pluribus, an AI that specializes in six player no-limit Texas hold’em poker. While Nash equilibrium isn’t applicable at every step of the AI, game theory still proves essential to its success.
In games such as checkers, a zero-sum game, AI’s have achieved great success against human players by approximating Nash equilibria strategies. Of course, compared to computers, humans are limited computationally, so have to often defer to other strategies. A more human approach to a game is to evaluate the weaknesses of an opponent and, combined with judgement and a bit of luck, use this to succeed. It would thus make sense that a successful artificial AI for poker might both approximate Nash equilibria strategies and evaluate weaknesses of opponents. However, such a solution would involve algorithms that have proven to be elusive. Thankfully, AI’s have still managed to prevail.
Pluribus employs a fixed strategy so the tendencies of opponents are, interestingly enough, entirely ignored. Due to the difficulty of building an efficient algorithm for finding Nash equilibria, Pluribus builds its strategy by playing against copies against itself continually, iterating upon its own knowledge. With Pluribus, an algorithm known as counterfactual regret (CFR) minimization guarantees that a Nash equilibrium strategy is eventually converged upon as the average strategy after over time.
With these aforementioned strategies and components, Pluribus has managed to defeat professionals playing for monetary prizes of thousands of dollars. Pluribus demonstrates both the incredible power and the limitations of Nash equilibria. While Nash equilibria prove essential to the AI, there’s also much more that makes Pluribus successful. Programs playing poker are fine and all, but it’s the implications of the existence of Pluribus that suggest a bright, game theory fueled future for AI.
Social Media and Social Networks
Social media is often though to increase social networks because of the access it gives to reach new people without barriers such as distance. However, it has been shown that social networking with social media as compared to social networking in person and with traditional methods had surprising results. evolutionary psychologist Robin Ian MacDonald Dubar studied this relationship and found that online social networking does not expand the size of our genuine friendships. These relationships that are gained through social media are thought of as “loosely defined acquaintances”. It has been shown that social media helps to maintain already strong relationships between friends, decreasing lost strong ties for reasons such as distance.
We can look at these results in terms of strong and weak ties between friends and relatives. The social networking can help individuals to gain a lot of weak ties with people they don’t yet know. However, it is not adding any new strong ties. It does help to maintain already strong ties. If we think about the impact of a bridge, gaining weak ties through social media could help to achieve access to another friend group or another social network. If we think about the strong triadic closure property, a lot of connections satisfying the property would be newly formed. Since the newly formed relationships are usually that of weak ties, we would not see a lot of new connection between friends of friends. We would see the continuation of strong ties from already established friendships, and therefore the satisfaction of this principle with people who were strongly connected without the help of social media.
Overall the intent of social media helping people gain many strong relationships may not have been achieved, however there are other benefits to online social networking.
https://www.techwalla.com/articles/the-positive-and-negative-effects-of-social-networking
Game Theory: David Cummings’ approach
https://www.independent.co.uk/voices/brexit-dominic-cummings-game-theory-boris-johnson-parliament-supreme-court-a9106926.html
In this article we learn about David Cummings, Boris Johnson’s advisor, and his application of game theory to politics. The article discusses how Cummings is interested in game theory; however, his current decisions might not be correct. The obvious two players of the game from Cummings perspective would be the EU and the UK.
The article provides an example of Cummings using game theory when in August Johnson started to plan to suspend parliament by the October 31st deal or no deal Brexit date. The goal here would be to push the EU to provide a better or at least altered deal.
Nonetheless, the author explains how Cummings might have not understood all players involved in this “game” and thus his decisions are not correctly based on game theory. The third player in this game according to the author are the British MP’s. This became apparent when the MP’s successfully stopped Johnson’s ability to carry out his do or die Brexit pledge.
Cummings and Johnson underestimated both the speed and decisiveness of the members of Parliament. They provided one week for the MP’s to act before the shut down. One of the most important parts of Game Theory is to look at all possible outcomes and then go on to make the best decision. In this situation Cummings and Johnson did not consider how quickly the MP’s could act and their negligence lead to the current outcome.
As we have discussed in class only through understanding all outcomes can one accurately choose the outcome that will lead to their best decision. In this situation Cummings and Johnson thought they were forcing the EU into a better “Brexit” deal; however, because they did not see the whole picture, which in this case included the MP’s, they could not accurately choose a path.
Using Prisoner’s Dilemna To Help Model Cooperation
https://www.sciencedirect.com/science/article/pii/S009630031830554X#keys0001
Networks that are seemingly unrelated may actually change the behavior between members of their respective networks. According to the academic paper, while networks themselves may not share any nodes or ties, they often bleed information to people that belong to other networks, altering their decision making. In order to generalize this idea for an experiment, the researchers needed a payoff matrix and a probability model. The payoff matrix is based on Prisoner’s Dilemma, which the payouts are constant except for the temptation to confess. The temptation to confess changes based on the amount of knowledge one person has over the other. This knowledge gap is known as asymmetric information; members of one network have more information than members of another network. The results of the experiment were that people are more likely to cooperate when information is more asymmetric.
At first I thought this result was counterintuitive, thinking that people with the same information would come to the exact same conclusion and cooperate. However, after conjuring up an example similar to that from lecture, it made sense to me how this could apply in the real world. Assume 100 cars want to get from point A to point B, and there are 2 paths, p1 and p2. p1 takes x/50 +0.1 hours and p2 normally takes 0.1 hours, but because there is a car crash on this road, it takes x/50 +0.1 hours to take p2. Assume that 50 cars have access to a radio station that tells every listener about this crash and the new amount of time it takes to travel along this road. This is a different network creating asymmetric information among the members of the traffic network. Without any information from the radio, everyone would choose to take p2. Assuming everyone who hears about the crash switches paths, the cars along both paths would be equal, and everyone would arrive at point B in 1.1 hours. This is the minimum travel time possible. If everyone had the same information, and everyone or no one decided to switch paths (the case of no asymmetric information), every car would arrive at B in 2.1 hours. There is no cooperation in this case. This is just one possible example where asymmetric information leads to cooperation.
Knowing Social Network and Using Health-Tracking Device Can Help Predict One’s Well-Being
According to the article, the researchers at the University of Notre Dame found that using a health-tracking device (namely Fitbit) and knowing one’s social network increases the predictability of one’s mental and physical health than using Fitbit alone.
For the study, the participants wore Fitbits to keep track of basic health data, including steps walked, hours of sleep, and heart rate, etc. and self-assessed about their stress level, happiness, and other mental qualities. In addition, the researchers analyzed participants’ social networks, so that they could find a correlation between the participants’ overall health and well-being and social networks. This involved machine learning and calculating metrics such as clustering coefficient and the number of “triangles” within the network. The study showed that there is a strong correlation between social network structures and heart rate, steps walked, etc.
This research is closely related to the topics that we learned in class; this research counted the number of triangles in the network and derived a clustering coefficient. This clustering coefficient would be valuable because it may give the researchers an insight on how extensive one’s social relationships are, since higher value of clustering coefficient would mean that one’s friends are more likely to know each other. This may help predict one’s social health and ultimately, one’s mental health and well-being.
I believe that this research may help improve the predictability of Fitbit and other devices on one’s well-being by encouraging the implementation of social networking function to the devices. However, I believe there might be a potential for creating unnecessary competition among friends — for example, on how much they have walked in each day — which may not help them on predicting their well-being.
YouTube recommendations and “children’s” videos
TED talk: https://www.youtube.com/watch?v=v9EKV2nSU8w
Toys Trek video: https://www.youtube.com/watch?v=odd04DhsnMM
When a user browses YouTube, YouTube will recommend videos that “go well” together. That is, it will recommend videos that are similar to each other, or it will sacrifice some of that similarity if the video is very popular. This is reasonable. If a viewer watches a video, they will probably like similar videos. And if a video is popular, more viewers are likely to enjoy it. Correspondingly, YouTube tends not to recommend videos that are unpopular or dissimilar to those the viewer watches. This is also reasonable. There is no point in recommending such videos if YouTube knows which videos will get them views.
Then, we can think of YouTube videos as a massive network, where videos are connected if they “go well” together. We can imagine YouTube’s video network as having massive clusters of similar topics. For example, videos on carpentry will probably be clustered together, videos on gaming will probably be clustered together, and so on.
But who decides who decides which videos “go well” together? On one hand, it is the viewers, since they determine the popularity. But when a video is first put onto YouTube, who decides? In this case, it is a complex algorithm that looks at the title, the description, the channel, and maybe even some of the video, and then says, “Yes, this goes well with carpentry.” or “Yes, this goes well with gaming.” Of course, the algorithm doesn’t actually have any idea what carpentry or video games are. It really just mashes the data it receives together and places the video “close” to videos it thinks are similar.
When I put it this way, it seems like a wonderful system. No human curation! No man-hours wasted! The recommendation engine JUST WORKS! But unfortunately, this engine had – or maybe still has – a major flaw.
This flaw is simple. You can trick the algorithm. Consider the cuckoo bird. The cuckoo bird lays its egg in other bird’s nests. The bird that owns the nest looks at the cuckoo egg and thinks, “It looks egg shaped and is in my nest, so it must be one of my eggs.” In this way the cuckoo bird can have offspring without ever having to care for them. In the same way, content creators can and do mash words, tags, and video content together, to trick the algorithm into thinking, “Ah, yes, this MUST belong in carpentry!” when in reality it doesn’t.
But where this is most prevalent is not carpentry. It is in children’s videos. Very, very young children. So you end up with titles like “NEW 101 SURPRISE EGG OPENING PAW PATROL MOANA COCO SHOPKINS PJ MASKS MICKEY DISNEY MLP MARVEL PEPPA” (This one is courtesy of the channel Toys Trek) which only serve to mangle the algorithm into planting this video firmly in the “kid’s” cluster of YouTube (as illustrated by the TED talk).
Which is honestly fine. The video itself is just someone opening eggs and other containers. But the problem arises when people use these same techniques but also begin to lace these videos with disturbing and highly inappropriate imagery, considering the target audience. These videos show things like animated gore, cartoon characters engaging in violence or being a victim of it, and characters in sexual situations (also shown in the TED talk). And because the algorithm has no real intelligence, it plops these videos firmly in the kid’s section.
For these content creators, they seem to have a clear formula. First, they load every aspect of their video and its posting with content that will signal to the algorithm that it belongs in the kid’s section. Second, they fill their video and posting with content suggestive or explicit of violent or sexual themes, which serves to “clickbait” kids into watching it. Third, they sit back and watch that sweet, sweet ad revenue roll in.
This phenomenon of disturbing “children’s” videos was termed “Elsagate”. Back when it was big, YouTube tried to clear out most of the inappropriate videos. And it seemed to have worked… mostly. Still, it serves as a word of caution: in a network, be careful of what connects to what.
Game Theory and Human-Robot Interactions
https://www.nature.com/articles/s42256-018-0010-3
In January, a research team led by the University of Sussex for the first time managed to use game theory and the concept of Nash equilibrium – the state that a system will reach given each “player” in the game attempting to use the best response strategies to each other – in programming robots that can interact better with humans. Previously, most such attempts focused either on complete control by the user, or a fixed pattern that cannot easily adapt to potential changing requirements and needs of the human. Game theory had been suggested prior as a framework to model interactions between the two entities, but the problem was that conventionally each partner requires complete knowledge of the dynamics and strategies used by the other. The University of Sussex team, however, managed to get around that limitation by using adaptive techniques in order to learn and model the human’s behavior as a type of controller, and then applying game theory and the robot’s knowledge of that control to compute a Nash equilibrium for the system. This allows for a stable interaction between the two “players,” enabling them to optimally work together on a task.
The published paper focused primarily on physical interaction between the human and the robot, such as in physical rehabilitation, where doing so could be most helpful. In the example mentioned in the paper, the researchers conducted an experiment to validate the procedure of arm reaching movements of patients. However, it seems very conceivable that the underlying principles behind the method could also be applied to social interaction between robots and humans. After all, human interactions often can be modeled using similar game theory techniques, since in many situations, much of what we do can be boiled down to finding the optimal response to the needs, questions, or statements of our counterpart. Extending this to a social context needs to connect further to the concept of networks, as the strategies of the human user that the robot would have to consider would depend on the state of the entire network, as well as how its behavior changes over time.
In short, this paper shows that the ideas of game theory have far broader-reaching applications than just to fields of economics and politics, the most commonly known examples. This application of game theory to robotics seems to have the potential to become an important player in the near future as technology and AI advance.
« go back — keep looking »