The Influence of Information Networks Over Voting Outcomes
http://news.mit.edu/2019/information-gerrymandering-influences-voters-0904
A recent study designed by MIT scholars aims to understand the influence that information networks have over individual decisions. In particular, the study examines the impact of political information on voter behavior, proving that different information networks can increase the chance of electoral deadlock or bias overall election outcomes in favor of one party. The results show that “information gerrymandering,” the act of establishing a political advantage for a particular group by manipulating voter information, can bias the outcome so that one party wins up to 60% of the time in simulated elections.
The study consisted of over 2,000 participants divided into two teams, meant to represent two opposing political parties. The participants played a voter game in different conditions, in response to continually updated polling data information. The study used payoffs as incentive to produce certain outcomes – a team would receive a large payoff if their own team received a super-majority share, a smaller payoff if the opposing team received the super-majority share, and zero payoff if neither team did. The data showed that participants voted for their own party when the polling data showed they had a chance of a super-majority, or if it showed a deadlock was likely. However, if the opposing party was likely to win, half the players would vote for the opposing team, and the other half would continue to vote for their own team.
The study manipulated information through information gerrymandering, in which some members of one team were “placed inside the other team’s echo chamber,” and through deploying online bots that strongly supported one team over the other. The results showed that election outcomes could be strongly biased based on how the polling information was distributed over the networks. When members of one team were led to believe that most of their team were voting for the opposing party, they often switched their votes to avoid a deadlock and receive a payoff.
The results of this study are consistent with the ideas of game theory that we covered in class. In this case, participants will choose their voting strategy based on the knowledge they are given. Their main concern is to avoid a deadlock in order to receive some sort of payoff – if the other team is likely to win, their best strategy is to vote for the other team. This shows that players choose their best strategy based on the information they are given for voting outcomes.
Game Theory Behind What We Eat
Presh Talwalkar believes that following health claims on food and drinks may not always be a good idea. Labels are, in fact, a “trend” in today’s diets. And though this may seem ridiculous, it’s actually true. He notes that a package of white rice has been labeled with “gluten-free”, “lactose free”, “low sodium”, and “vegan”, which rice naturally has. Why? People doubt things unless they’re told. If someone were vegan, they would be unsure if a product is vegan, unless it were labeled. He relates this to wealth, and how people show wealth — if you wear fancy clothes and have an expensive car, people will assume you are rich. In this sense, the “labels” are a way of making sure people see the health content of a product, and be able to compete with similar products, diving into the issue of people’s trust in words, because it removes doubt. And with this, Talwalkar wants us to have common sense regarding what is healthy, rather than relying on labels.
So, why is labeling still a trend? This question brings us into the concept of Game Theory. Much like the Prisoner’s Dilemma, let’s say we have two companies A and B, with a similar product. Both products are lactose free, low sodium, and vegan. They have the choice to label their product as such, or not at all. So what happens? If A decides not to label their product, and B does not, both have an equal chance of gaining a consumer. If A decides not to label and B does, B has a greater chance of gaining a consumer. Thus, in this case, the best response for B would be to label the product. If A decides to label their product, then B’s best response would also be to label their product, else A would have a higher chance of receiving a sale. And, this situation is the same for A. So, the dominant strategy for both companies ends up being to label their product, and thus, the reason why obvious labels have become a “trend” in foods.
How Facebook Dating Might Surpass Tinder
The article “Can Facebook Fix the Dating World Tinder Created?” from The Atlantic by Ashley Fetters brings up some interesting points about how the culture of dating nowadays has evolved. It has become a modern trend to find dates or partners through Tinder, Bumble, and other dating apps. Tinder now has a worldwide user base of around 50 million, and other dating apps have soared in popularity as well due to advancements in technology and acceptance of this way of dating. However, dating app users often complain about matching or going out on dates with people whom they know nothing about. If you think about it, users are judging a potential date’s character based on some profile pictures and a short bio. This creates a small possibility of having a previous connection or so-on-so “common ground.” That’s why Facebook wants to change that with their new matchmaking service called Facebook Dating, which would algorithmically connect singles by matching them based on their geography and shared interests, events, and groups. Facebook users can also opt in or opt out of matching with their Facebook friends’ Facebook friends since they don’t have to match with their own Facebook friends. Facebook conducted a survey that polled 3,000 Americans above the age of 18 of whom 40% felt that the available dating apps and sites weren’t meeting their needs. As well, the survey results showed that similar interests were the top-ranked trait most people were looking for in a partner, over looks and financial prospects. Although it may seem unreliable that people with shared interests are more likely to lead to dating, Facebook is aiming to create the experience of meeting someone in person virtually with Facebook Dating.
Whether or not Facebook Dating is just another way for Facebook to bounce back after having private Facebook data harvested by Cambridge Analytica, it is an interesting take on online dating. Would it be better than Tinder, or in other words, would it satisfy the needs of more people than Tinder has? Let’s look at it this way. For every common interest a Facebook user has with their own Facebook friend, they have a strong tie. Or if the user is close with that Facebook friend in general they have a strong tie. Many users are friends with people on Facebook with whom they have no common interests. For this user’s Facebook social network to satisfy the Strong Triadic Closure Principle, the strong ties the user has with their Facebook friends should theoretically also be connected with each other in some way. This means that the user will have a higher chance of sharing a common interest with one of their mutual Facebook friends. If the user has strong ties with two Facebook friends, there must be another tie that somehow connects the first friend to the third friend, whether the tie is weak or strong. On the other hand, Tinder users would have less strong ties with other Tinder users because they don’t have some sort of common ground to begin with. Through this logic, Facebook Dating would be able to increase the chances of Facebook users finding a date who are more suitable for them than Tinder has, that is, disregarding all other factors that could affect each user individually. Perhaps Facebook Dating will be the next most popular dating service, or it won’t but maybe people will realize that the conventional way of dating worked for a reason.
Game Theory & Why Kidnapping Rarely Pays
According to the research in Kidnap: Inside the Ransom Business, when foreigners are abducted in a kidnap prone area, they have a safe return rate of more than 97% of victims. The kidnapping of expatriates, tourists, and foreign firms’ local staff is discouraged by the unhurried, firm approach to negotiations and hostage incidents resolved by professional crisis responders. Additionally, the vast majority of professionally kidnapping negotiations conclude in less than a week and generally with ransoms that don’t break the bank. Granted that in these cases of abductions, kidnappers are usually “rational” criminals who would choose to take a smaller profit over having to face legal consequences, the reason why kidnapping rarely pays can be analyzed within the frameworks of game theory wherein such games, kidnappers would be prone to take a non-violent approach and be willing to establish negotiations of ransoms that facilitate swift and reliable release.
The game between kidnappers and negotiators representing the victim can be represented in the figure above. As kidnappers do not want to be held legally responsible, they would be likely to prefer to free the victim whether negotiators pay or decide to hold out. Even in the case where negotiators hold out, kidnappers would be “better off” to not kill the victim as they face greater risk of being punished by the law if they choose to kill. Freeing the victim would likely be the dominant strategy for kidnappers. On the other hand, the outcome for negotiators would be better (or less terrible) if negotiators choose to hold out and negotiate for the ransom rather than paying immediately and hence, holding out is the negotiators’ dominant strategy. In real life, there exists the risk of the kidnappers going rogue and killing the victim if their demands are not satisfied so the theoretical Nash equilibrium of negotiators not paying and kidnappers freeing the victim is very unlikely to happen. However, the analysis of this game should establish the intuition for kidnappers to refrain from violence and for negotiators to barter the kidnappers down during negotiations. This kind of intuition steers negotiations towards an outcome where kidnappers would leave the victim unharmed if a reasonable amount of ransom (instead of the usually astronomical first ransom demand) is offered. It is usually the case that depending on how negotiators respond, kidnappers will either revise the ransom upwards or downwards. If the negotiator appears to be willing and able to pay a greater sum of money for the victim, kidnappers would have the leverage to ask for more. It is helpful to think of this with economic cost-benefit reasoning: kidnappers will release the victim when the cost of holding on to the hostage exceeds what they expect to gain from the next ransom increment. Thereby, it is best, when playing the game of ransom, to be conduct negotiations with a firm, calm and cautious approach to bring the victim back safely with a price that ideally would only cover the costs of staging the kidnap.
Source: https://theconversation.com/inside-the-ransom-business-why-kidnapping-rarely-pays-110678
How Athlete Doping Relates to Game Theory
https://mathsection.com/a-game-theoretic-analysis-of-doping/
Game theory can be applied to many real-life situations. This article analyzes the incentives of athletes to use performance-enhancing drugs in the Olympic Games. Doping was prevalent in the 2004 Athens Olympic Games, and it has been branded one of the dirtiest games ever. So why exactly did these highly skilled athletes feel the need to cheat? To analyze this, first we need to make some assumptions. Let’s assume we have two athletes of equal ability which are competing against each other in the Olympics. The athletes have two strategic choices—to dope or not to dope. We assume the player that takes the drugs wins the race of certainty. Thus, if both athletes take the drug, they are once again equal in ability leveling the playing field. We assume both athletes to start simultaneously before the competition whether to dope or not to dope. Finally, we assume perfect information and rationality. We’ve learned from the class that a key aspect of a game is the payoff a player can potentially receive depending on their choice and their opponent choice’s strategy. The numbers in the payoff matrix below describe how happy the athletes are with a given scenario.
| A\B | Dope | Do not dope |
| Dope | (2,2) | (4,1) |
| Do not dope | (1,4) | (3,3) |
If athlete A one chooses to dope, athlete B can dope with a payoff of 2 or not dope but would receive 1, so he would choose to dope. If athlete A chooses not to dope, athlete B would choose to dope because 4 is larger than the payoff of a 3 if he were to choose not to dope. If athlete B chooses to dope, so will athlete A as 2 is larger than the payoff of 1 if he were to choose not to dope. If athlete B chooses not to dope, athlete A will still choose to dope because the payoff of 4 is larger than the payoff of not doping of 3. Therefore, (dope, dope) is a Nash Equilibrium as neither athlete can gain by unilateral change of strategy.
Individually, both athletes are worse off than if they were to both not dope. This is an example of the tragedy of the commons which we’ve talked about in class. Because the payoff for doping is greater than the payoff for not doping, despite what the other athletes chooses, each athlete has a strictly dominated action of doping. From the lecture, we know that this type of game is referred to as the prisoner’s dilemma game. The optimal outcome for both athletes and for society would be if neither used performance-enhancing drugs. To achieve this, the payoffs need to be changed so that there wasn’t an incentive to dope if another athlete was to dope. Hence, to eradicate doping from sports and help athletes get to the optimal scenario where no one is doping to boost performance, IAAF could achieve it by increasing the probability of being caught or doing more extensive checks covering a wider range of athletes and increasing the length of the ban from sports or by imprisonment.
Prediction of Character Through Networks
referenced articles:
(1) https://www.sciencenewsforstudents.org/article/social-networks-can-learn-about-you-through-your-friends
(2) https://journals.sagepub.com/doi/full/10.1177/2056305116664219
This article (1) focuses on the impacts of social networks and how anyone can find your personal information, whether you personally uploaded it or not, through friend connections. However, the paragraph that interested me the most was about how David Garcia, who studies how people interact with social networks at the Complexity Science Hub Vienna in Austria, discovered that he was able to predict the characteristics of peoples’ friends based on their characteristics.
“Most people don’t have a random assortment of friends. Married people tend to be friends with other married people, for example. But people also have connections that complicate the ability to predict who is connected to whom. People who identified as gay men were more likely to be friends with other gay men. But gay men were also likely to be friends with women. Straight women were more likely to be friends with men.”
Through these observations, Garcia was able to predict the characteristics of certain peoples’ friends, even if those friends had never had accounts on Friendster, the outdated social networking site he researched. “He could predict things like whether someone was married, or whether they identified as gay. The more people in the social network who shared their own personal information, the more information the network received about their contacts. And its predictions about people not on the network got better, too.”
This article, specifically this section, emphasizes the concepts of the properties we learned in lecture. People are more likely to be friends with whom they find similarities in, a concept discussed in the Strong Triadic Closure Principle. The reading also adds depth to the ideas Granovetter discussed in his questioning of “the strength of weak ties” — many people ‘friend’ other people on social networks even if for reasons that don’t imply friendship, as discussed in (2). An anonymous participant in the study discussed in (2) even said, “Thank you for making me realize how little I care about my friends on Facebook.” And although Granovetter most likely wasn’t speaking of the impact these weak ties may have when he mentioned their ‘strength,’ we can see another meaning to it in today’s society, filled with networking: even the weakest of ties can reveal a plethora of information about the nodes in a network. A simple click of the ‘accept friend request’ button can give access to personal information and can unsheathe what we once believed was pure privacy. The strength of weak ties, truly, lies in the lack of privacy in today’s world wide web.
Are You Your Friends’ Friend?
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0151588
A 2016 study has concluded that only about half of perceived friendships are mutual, where both individuals consider the other a good friend. This is an issue that is widespread and affects everyone, as it relies on the foundational concept of friendship, something everyone should be familiar with. This disheartening study relates to directional graphs (which can indicate a one-sided relationship in the case where it is not mutual), which are likely weak ties as well.
The study asked respondents to score the strength, 0 through 5, of their friendship with another candidate in the study, and predict how strongly the other candidate would rate the relationship on the same scale. Though almost all (94%) of candidates expected their relationships to have reciprocal strength, only about half (45%) of the relationships similarly (within 2 points) scored each other. The researchers modeled the friendship network with dashed and solid ties, where dashed indicated uneven relationships while solid referred to reciprocal ones. Interestingly, the graph produced from the data of the study shows clumps of close (reciprocal) bonds with many weak (nonreciprocal) bonds bridging out to other components within the network many times over, indicating people may overestimate how friendly they are with people outside their direct friend group or family (there are a lot of fascinating graphics in the article, definitely take a look!).
This material raises questions about the concept of strong and weak ties that we have covered in class. Is it as simple as those two categories, or should one consider whether or not persons A and B are on the same page with their relationship? Many strong ties may, in fact, be lopsided relationships. This further brings into question the Strong Triadic Closure Property and how well that truly holds considering how many relationships lack equality in strength from either party.
Hiring a Friend — A Structural Balance Analysis
https://www.forbes.com/sites/jeffhyman/2019/01/16/friends/#5298f6a84b23
Professor Benson mentioned in class that more people found jobs from acquaintances(friends of friends) instead of good friends. I was very interested in verifying whether this was true, so I looked up some articles about the subject. Turns out, there were many articles that advised against hiring friends, mainly because they changed the workplace dynamic, and very easily lead to conflicts and mistrust. As mentioned in the Forbes article, “People on your team will be wary at even the sniff of favoritism toward your friend. They fear that your friend will be your ‘eyes and ears’ in the workplace. Some may try to curry favor with the person, while others may feel threatened and keep a cool distance.”
While the article itself doesn’t spend a lot of time talking about the implications of such a workspace dynamic, I found that it the workplace could be easily understood using the structural balance property — or, more specifically, how the structural balance changes because of the addition of another nodes or individual to the network. Supposedly, a workplace should be (hopefully) a structurally balanced network. If all goes well, the workplace should only have positive edges, and everybody should like each other; of course, it is also not uncommon for a workplace to have two factions opposing each other (which would still be a structurally balanced network by the balance theory).
Introducing a new individual or node into the network changes the structural balance of the workplace. You would be forming a positive bond with this new hire, since you are great friends, and you will probably have positive edges with some of the original hires because you work closely with them. However, because of the risk of “favoritism”, the new hire and the original hires are more likely to form negative edges because of mistrust. This mistrust means its more likely for the new hire to form negative edges with original employees. This creates an unbalanced structure within the workplace, which leads to the poor morale and increased tensions mentioned in the article.
Of course, the article never said that it is impossible to maintain friendships after hiring close friends, but as the structural balance property shows, it is very difficult — the new hire will have to form positive edges (or befriend) all of your friends in the workplace, and form negative edges (or dislike) all of the people you dislike. While possible, this doesn’t seem very likely to happen — so maybe it is a good idea to keep work and private life separate after all.
Here’s how social media could threaten democracy — even without the help of Russians
“Here’s how social media could threaten democracy — even without the help of Russians” discusses the impact of the rapid spread of information on social media. The article explains a research study that tested how online social networks impact decision making. 2,500 people were split into groups of 24. Each group was split into a yellow and a purple team. If your team wins a supermajority of a vote, each member gets $2 and the other team’s members get 50 cents. If no team has a supermajority, no one earns money. Therefore, opponents are encouraged to compromise. The experimenters also designed different games that skewed the network in order to make each subgroup believe that either the yellow or purple team was at a disadvantage by creating an uneven yellow:purple ratio within subgroups. Through analyzing the behavior of the participants, researches concluded that when teams were split 50-50 purple and yellow, participants were more likely to compromise. Meanwhile, when one team held a majority in a subgroup, those team members were more likely to hold an extreme view. This concept also applies to online networks, especially regarding political parties.
Therefore, if you are a democrat and are exposed to extremely liberal content online, you may transform your views and become a more extreme liberal. Meanwhile, if you are exposed to diverse, well-rounded media, you may be more likely to understand different political views. Online networks, however, perpetuate one sided media consumption and contribute to political polarization.
This type of political network may be related to the Structural Balance Property. When a group of people encounters a group they dislike, structural balance exists. In the skewed version of the game where yellow and purple players fail to negotiate, yellow players have positive feelings towards one another and purple players have positive feelings towards one another. However, members of opposite color teams will likely have negative feelings towards one. These opinions are therefore unlikely to change, as when there is tension during the game, players on the same team will display negative feelings towards members of the other team.
This study also relates to the University of Michigan (2004) study we discussed in class which analyzes networks of people reading political blogs. People who read liberal blogs are more likely to be led to read more liberal blogs in the future. The same holds true for conservative blogs. There is a resulting echo chamber effect, as when these networks are formed, there are two separate clusters. Few people are reading blogs that represent different ends on the political spectrum.
What game theory tells us about nuclear war with North Korea
Link: https://www.washingtonpost.com/news/wonk/wp/2017/08/16/what-game-theory-tells-us-about-nuclear-war-with-north-korea/?noredirect=on
This article was about how the rising tensions between the USA and North Korea have reached a point where the next move can seem unpredictable. It is hard to guess what might happen next, and this is because there have been so few cases of a nuclear war, and without much background to these types of wars, analysts have trouble thinking about how to stop them. However, with Game Theory, the nuclear war tensions can be better understood, and there has been a history of using Game Theory to think about military issues.
Game Theory is something that we have discussed in class, and the article also described it as using mathematical reasoning to better understand two “players”, or sides, as well as their strategies. In this situation, the two sides would be the USA and North Korea, or even Trump and Kim Jong Un. The article mentioned that this would be an example of a Prisoner’s Dilemma, which is when both sides have an incentive to betray the other, but it would best if both of them just stayed silent. With North Korea and the USA, it would be a repeated Prisoner’s Dilemma, and even though each has an incentive to attack, because of possible retaliations and future wars, it is best if they both stay put and do not attack. There may not exactly be a Best Response in this situation, because Best Responses require that you know what the other side is going to do, and that is simply unclear here.
« go back — keep looking »