Categories
Uncategorized

How Prisoner’s Dilemma and Game Theory can explain people’s behaviour during the COVID-19 lockdown.

In a recent lecture, we were introduced to the topic of game theory and a specific type of game: “The Prisoner’s Dilemma”. In this game, the strategies that the rational players will play is not the strategy that gives the greatest payoff for both players. The article “Game Theory Explains the Pandemic”, the author Malini Sharma likens the choice to social distance during the pandemic to the Prisoner’s Dilemma.

While social distancing should be practiced with all members in a community, to demonstrate how social distancing can be a Prisoner’s Dilemma game, she sets up a two-player game and each player has two strategies: strategy “C”, which is the choice for the player to maintain social distancing and strategy “D”, which is the choice for the player to not social distance, i.e. go outside of their home. The payoff matrix in the article looks like this:

We can see that this is very similar to the payoff matrix of the Prisoner’s Dilemma as shown in class, and that the best strategy for the two players would be (C, C) with payoff (5, 5). However we can see that both players the strategy they will choose is “D”, as that is the dominant strategy for both players regardless of what the other plays. This means it is “best” for both players to not socially distance which is strategy (D, D) with payoff (1, 1). The author explains why this could possibly happen. If player A stays inside and socially distances but player B goes out, the payoff for the person who goes out is 8 and the one that doesn’t gets 0 payoff, she states that “player A will get a small benefit (8, 0) but at the cost to others in the society”. Although she doesn’t describe exactly what the payoffs are, it makes sense. The person who goes out gets to go to work and make money, or go out and enjoy themselves (which is the payoff). Meanwhile the person who doesn’t go out doesn’t really get the benefit of social distancing because of the other player breaking the rules and they do not get to go out, possibly meaning they are unable to work, or maybe their mental state is not as good due to the isolation, hence the lower payoff of 0. This means if both players choose strategy D, they get a payoff of being able to go out, but it is severely decreased as the risk of infection gets higher.

Overall, game theory could possibly explain the current state of Ontario’s COVID-19 situation. From an article from the CBC many of the people who contracted COVID are younger people and they speculate that the reasons why they break social distancing are:

  1. That they do not take the pandemic as seriously
  2. They need to go to work

Rewording this in a game theory way, we can say that for the first reason, the decrease in payoff to not social distance is not enough to deter people. The second reason would be that going to work and risking COVID-19, gives a greater payoff than staying home. Finally, the question now is how could Ontario increase the numbers of younger people social distancing. With the game theory model above they have two options:

  1. Decrease payoff of breaking social distancing rules (e.g. fines)
  2. Or increase payoff of staying home and following social distancing (e.g. financial support like CERB)

Sources:

Categories
Uncategorized

Formation of Small-world Networks on Twitter

In lecture, we learned about “small-world” networks which are graphs characterized by high clustering and short paths. We were also shown how to construct them using Watts & Strogatz model. However, I wanted to see an example of a small-world network occurring in real life. After some searching, I came across a journal article “Local Interactions and the Emergence of a Twitter Small-World Network”, written by Eugene Ch’ng.

The paper was published in 2015 and the researcher analyzed Twitter activity relating to one of the suspects of Boston Marathon bombing back in 2013, Dzhokhar Tsarnaev. There was a small group of people who insisted on his innocence under the hashtag #FreeJahar which caused a lot of controversy. So, the researcher proceeded to collect data from Twitter over the course of five hours using #FreeJahar and related hashtags and words. Using the data, he constructed a directed graph which had two types of nodes: a user node and tweet node. If edges went from a user node to a tweet node, that meant the user published that tweet and if an edge went from a tweet node to user node, it meant that the tweet mentioned that user.

The researcher showed that in the time period of five hours a small-world network formed from the dataset. The large cluster at the top represents interactions between users relating directly to #FreeJahar. This cluster includes both conversations between the supporters and the opposers of the movement. We can see there are smaller clusters forming under the large cluster. These smaller clusters were centered around activity from news channels. As time went on the clusters started to become more clustered and more connections began to form between each cluster. The data below the small-world network are various isolated interactions relating to #FreeJahar.

The properties of the networks the researcher found was interesting. The left graph compares nodes by their degrees. The nodes with the highest degrees are the news channels, which makes sense as people typically retweet news stories. The right graph shows the betweenness of the nodes, and the nodes in the main cluster had the highest betweenness. Ch’ng concludes that since the nodes in the main cluster have the highest betweenness, these nodes are the ones creating and maintaining the small-world network. This was confirmed when eventually, Twitter accounts belonging to the main cluster got suspended and he saw that the small-world network broke apart. He also noted that in the right graph, there is two ‘sub-clusters’ inside the big cluster, one denoting the supporters (red) and the other the opposers (green). Formation of this big cluster happened because the supporters and opposers would interact in the form of arguments.

In conclusion, the author of the paper states that the formation of small-world networks on Twitter during a news event could help people that are supporting or opposing a cause. For example, if a Twitter account is spreading misinformation, the suspension of that account can contribute to the fragmentation of the small-world community. This causes path lengths between users to become longer and make it harder for misinformation to spread. Although this article was published in 2015, I thought it is very relevant today, as many crises today are being brought to our attention today through the use of social media.

Source: https://arxiv.org/ftp/arxiv/papers/1508/1508.03594.pdf