Theme:
Problem Articulation:
Theme:
Underlying Problem:
What is the problem: Social media, which was designed to create a way to connect with society, is creating polarization and echo chambers among its users through their algorithms and how it helps the social media companies to generate more revenue.
Why it is a problem: Due to polarization and confirmation bias, the propagation of misinformation and misleading information has gone so pervasive. Users instead of believing truth fake news spreading on social media misleads them in other directions.
Key Variables:
Users engagement in politics (Year 2014)
The above bar chart depicts a survey done on American Public in 2014. The survey done on the total population is 10,013. Among all, they were divided into various categories. It can be seen that Gen X were 11% consistently liberal, 23% moderately liberal, 42% mixed, and 7% consistently consistent. Similarly, other categories were seen having different percentages of opinions.
Spending by political parties on advertisements (from year 2008-2020)
The graph depicts spending by political advertisements from 2008-2020. It can be observed that from 2008-2011 there was almost negligible spending by political parties on advertisements. After that, it rose to 159.21m $ in 2012 but it decreased to 71.16 m$ and suddenly it observed a spike to 1415.38m$ in 2016. It kept fluctuating but it is 2847.16 m$ in 2020 which is at its peak in the given time interval.
Comparison of other news media with social media usage according to user demographics (year 2016)
In the bar chart, it can be seen that compared to traditional news media systems such as news websites, cable tv, local tv and network nightly news social media usage is the highest among young adult users (age 18-29). Similarly, use of cable tv is the highest among old age people of age between 50-64 years. Also, the use of cable tv is the highest among adults aged 30-49 years.
Personal Discernment(Year 2016)
The bar graph above illustrates the views based on four different categories such as algorithm, subscriptions, usage and combinations. It can be observed that content based on algorithms states the highest views as 39.7%, based on subscriptions it is 27.4%, based on usage it is 10.1% and based on combinations of all it is 22.8%. Hence a sort of personal discernment can be observed.
Potential users usage of social media(Year 2020,2021)
The bar chart above illustrates the percentage of adults who use social media. In 2020, it can be seen that 23% of adults use social media often, 30% sometimes, 18% rarely, 21% never and 7% didnt get digital news. In the next year, it can be observed that users decreased from 23% to 19% who often use social media. Users increased by 1% who sometimes and rarely use social media. Also, users increased as 24% who never used social media and 9 % who didnt get digital news.
Attractiveness of the social media platform(year 2021)
It can be observed from the bar chart above that there are multiple social media platforms such as Quora, Reddit, Twitter, Pinterest, Snapchat, Telegram, Tik Tok, WeChat, Instagram, Whatsapp and Facebook.
The lowest used social media platform was Quora which has 300 million followers. Whatsapp has the third highest number of users as of 2000, Youtube remaining second highest with 2291 million followers. Facebook is the most attractive platform taking the lead with 2895 million followers.
Eco Chamber ( Year 2021)
The bar graph illustrates the consumption of news media feeds from four different categories such as posts from friends and people followed, posts from groups joined, posts from pages followed and unconnected posts. It can be seen that posts from friends and people followed remain the highest at 54.4% forming an echo chamber. Whereas posts from group joined and posts from pages followed remain at 17.8% and 14.9% respectively. In contrast, unconnected posts had views of just 11.7%.
Time Horizon:
For our model, we have taken the time horizon from the year 1800- 2000.
Since 1800, the polarization has increased from 0.5 to 0.7 and it has suddenly come to drop in the year 1820. From 1820, it has increased gradually and it can be seen that in the decade of 1900-1910 it is the highest.
In contrast to years between 1930-1980 it suddenly came to a drop and again it increased in the year 2000 due to increase in traditional news media systems.
Hence, we have tried to depict that even in the absence of social media platforms cases of polarization can be observed and the future predicts that it will keep on increasing in upcoming years.
Dynamic Problem Definition (reference modes):
In recent years polarization has increased due to increased usage of social media. In contrast to previous traditional systems polarization has increased tremendously due to extreme usage of social media.
This gives the user access to a wide range of information. However, because of the algorithms utilized by social media firms to give individualized material to each user, it has resulted in confirmation bias among its users. As a result, there is a bias toward one side of the political spectrum among the general public.
However, due to algorithm pre-selection of information and partisan news , dis-conforming news attitudes have increased and people have come to know about other information which contrasts their pre-existing beliefs. This results in lesser use of social media and thus polarization decreases. Hence, it can be seen that the polarization curve increases within a time period and gradually decreases due to contrasting information on social media.
Formulation of Dynamic Hypothesis:
Initial Hypothesis Generation:
In todays era, the effect of polarization and confirmation bias has increased due to exposure to social media. Users have unprecedented access to a wide range of perspective. This provides its user with a wide spectrum of information. However, it has resulted in confirmation bias amongst its users due to the algorithms used by social media companies which were designed to provide personalized content to each user. The result of this is seen amongst the population in terms of bias towards one side of the political spectrum. One of the examples of that could be the use of how social media has shown an increase in polarization as compared to traditional news media systems.
After certain experiments which included partisan news and non-partisan news it was found out that consumption of either attitude-confirming and attitude-disconfirming news caused attitudes to polarize and depolarize respectively. In a study it was found out that Facebook pre-selects information that confirms existing beliefs named filter bubbles and when people self-select the news feed based on their pre-existing beliefs which cause exposure to attitude confirming content.
Current scenario: In 2019, Facebook has announced new plans for making news feed even more suitable for personalization, it can cause problems as it would lead people towards polarizing attitudes to extremes. Also, if the trust in the presented news increased the polarization measure also increased leading to attitude confirming news.
Future scenario: The polarization and confirmation will continue to increase at extremes as online usage of social media platforms such as Facebook increases. As per current data in 2021, total population using social media has increased by 40% as compared to 2001. In the future, it is expected to grow by 80% and it will continue to mislead people because of their echo chamber rather than lead to information.
Endogenous Focus:
Endogenous Variable:
The variables that are included in the system are called endogenous variables and its relationship can be determined by other variables within the model. In other words, it can be stated as dependent variable. They are listed below:
Use of social media (I.e. Amount of time a person spends on social media)
Investment in social media Revenue
Brand Value
Total social media platform participation
Advertisements
Like, Share, Views
Attitude Disconfirming News
Eco Chamber (i.e., An environment in which peoples beliefs coincide with their own)
Social media trust
Potential users usage of social media
Polarization (i.e., Opinions of two contrasting groups)
Targeted content
Motivation to engage with voters via social media
Spending by political campaigns
Political ads on social media
View and share of political contents
Behavioral data gathered by social media platform
Tailored content to each user
Exogenous Variable:
Exogenous variables are those variables that are not affected by other variables in the system. They are listed below.
Stakeholder
Click baits (i.e. frequent clicks)
Excluded variable:
Excluded variables can be called as those variables that were not retained in the final model
Internet Access
Mapping
Sub system Diagram
In our subsystem diagram, we have focused on political parties and political engagement on social media. It can be seen that political parties take user data from social media such as Facebook and twitter. It invests revenue in social media to gain popularity. Users perceive news from traditional media such as cable news and radio stations and increase TRP. Due to political polarization misperception increases among users. Users are then biased in political engagement due to filtered content presented by social media. Political parties post advertisements and surveys on social media to gain favored decisions from the users. Due to selective content shown on social media, it increases users political engagement on social media.
Model Boundary Diagram
Endogenous Variable
Exogenous Variable Excluded
Investment in social media Stakeholder Internet access
Social Media Revenue Click baits
Brand Value
Total Social Media Platform
Participation
Advertisements
Like, share, Views
Attitude Disconfirming News
Eco chamber
Social Media Trust
Potential Users Usage of social media
polarization
Targeted Content
Motivation to engage with voters via
Social media Spending by political campaigns Political ads on social media View and share of political contents Behavioural data gathered by social media platform Tailored content to each user User engagement in politics through social media Confirmation Bias Effect of rumors Engagement Outside of social media Personal Discernment Interaction with people of different political belief Marketing on social media platform Attractiveness of social media platform Demand for independent journalism Misinformation Use of social media Spending by political parties for campaigning Social media companys revenue Number of Users Causal Loop Diagram
Revenue Generation Model
Explanation: When Investment in social media increases by stakeholders then social media revenue increases which leads to increase in brand value with some delay which in turn increases total social media platform participation which is followed by likes, shares and comments.
It increases advertisements which again increases social media revenue. More usage of social media results in eco-chamber which leads to confirmation bias and then polarization which increases targeted content in which people come around the news which discomforts their pre-existing beliefs and gradually social media trust decreases which results in decrease of brand value and total social media participation. Hence, it is a balanced loop.
Social Media and Political Engagement:
Explanation: Click baits increases view and share of political contents which increases motivation to engage with voters which increases spending by political campaigns which increases political ads on social media which makes it a reinforcement loop called engaging with voters on social media. Similarly, political unification is a reinforcing loop. In other scenarios, engagement outside of social media increases interaction with people of different political beliefs which increases personal discernment which leads to increase in effect of rumors which leads to increase in confirmation bias and user engagement in politics which reduces engagement outside of social media. Hence this forms a balanced loop.
Social Media Revenue and Voters Political Engagement on social media
Explanation: Social media revenue increases marketing of social media platforms which makes social media platforms more attractive which increases the number of users which increases use of social media which increases spending by political parties for campaigning. This makes a reinforcement loop called social media revenue loop. Similarly, political engagement on social media is a reinforcing loop. Due to political ads on social media, it results in spread of misinformation which increases demand for independent journalism which leads to decrease in attractiveness of the social media platform. This makes a balanced loop called the effect of misinformation on social media.
Policy Structure Diagram for Revenue Generation model:
Policy Structure Diagram for Polarization and Confirmation bias:
Stock and Flow Diagram
A.STock and Flow for revenue generation model
Specifications:
Advertisement in social media: With the posting of advertisement on social media, the participation of user on social media increases
Targeted Content: Based on targeted to specific audience according to their age,like and dislike,behavior,religion,backgroundLike, view and share: More response towards the post increases more containts.
Social media Trust: Potential user`s usage of social build communication and informational nature between users and company will improve trust on social media.
Advertisements: It controls flow going towards the Social Media Revenue stock defined as Advertisements.
Total Social Media Participation: Contribution on social media creates more sharing information between users such as likes, views and shares.
Potential user`s usages: It shows the behavioral date towards the particular area like brands, parties, companies on the social media.
Estimation:
Advertisement in social media = 0.1
Targetted Content = 0.05
Total Social Media Participation = Advertisement in social media+Targetted ContentLike, view and share = Total Social Media Participation+RAMP((Social media trust-Total Social Media Participation)/(Start time-End time), End time, Start time)Social media trust = Potential user's usage of social media
Potential user's usage of social media = Total Social Media Participation
Advertisements = "Like, view and share"*Social Media Revenue
Social Media Revenue = +Advertisements
Testing of revenue generation model:
Comparison to reference modes: Revenue generation model describes the revenue of social media companies through political ads, which aligns with the data shown in the diagram under key variables about spending by political parties on advertisements. Starting from the year 2008 political campaigns have heavily invested in social media companies to reach out to the mass and seems to grow in future.
Robustness under extreme conditions: This growth shows the same behavior under extreme values of the variables.
B. stock and flow of Confirmation Bias and Polarization
Specifications:
Input function: Effect of rumors and personal discernment are input functions of this stock and flow.
Confirmation Bias Rate: Personalized content on social media and input function will create more confirmation bias rate. This rate of flow is going towards Confirmation bias stock.
User Engagement in Politics through Social Media: Users involvement in the way of personal opinions and their likes and dislikes based on current events and situations. So, this will increase more user engagement in politics through social media.
Polarization Rate: More involvement on social media increases more polarization rate.
View and share of political content : Rate of change on polarization increases views and shares of political content. Users will share more content according to their choice. This will impact on Social media campaigns.
Spending by political parties on campaigns: Social media campaign rate flowing headed of spending by political parties on campaign stock.
Estimation:
Personal Discernment = 0.95
Effect of Rumors = 1-Personal Discernment
Input Function = (Effect of Rumors * RAMP (Time,INITIAL TIME,FINAL TIME)) / Personal Discernment
Confirmation Bias Rate = Input Function*Personalized Content
Personalized Content = 0.Confirmation Bias = Confirmation Bias Rate-Polarization Rate*Confirmation Bias
Polarization Rate = Confirmation Bias * Polarization Fraction
User engagement in politcs through Social Media Rate = Polarization Rate
User Engagement in Politics Through Social Media = User engagement in politics through Social Media Rate
Polarizations = Polarization Rate-View and Share of Political Contents * Polarizations
View and Share of Political Content Fraction = 0.25
View and Share of Political Contents = Polarizations * View and Share of Political Content Fraction
Social Media Campaing Rate = View and Share of Political Contents
Spending by Poltical Parties on Campaign = Social Media Campaing Rate
Testing Stock and Flow:
Comparison to reference modes: The results from the model shows that confirmation bias and polarization rises together, however the polarization rate would not be equivalent to the confirmation bias rate. Primary reason behind this is often the awareness about social media and its effect on the mental health of an individual. Moreover, events at the grassroot levels also decreases polarization among people.
B. Stock and Flow of Polarization and Revenue
Specification :View like and share of political views: Three major factors as mentioned in loop engagement with voters via social media, spending by political campaigns, Political ads on social media increases more views and likes as stated by behavioral data of users.
Polarization rate: More weightage of advertisement and political views and shares will increase polarization rate.
User perception: It depends on polarization rate, total population and misinformation data.
Misinformation : Polarization will create more misinformation. Some other variables are also responsible for spreading misinformation like fake news, rate of content sharing. It also demands for independent journalism to prevent wrong information.
Revenue: More misinformation increases more revenue and this will create brand value.
Estimation:
Polarization = +increase-Misinformation
Advertisement = 1
Motivation to engage with voters via social media = View like and share of political views
spending by political campaigns = Motivation to engage with voters via social media
political ads on social media = spending by political campaigns
View like and share of political views = Behavioral data
Increase = Advertisement*Polarization*(1-Polarization/(View like and share of political views+Polarization))
user perception = Total Population+increase-MisinformationMisinformation = Rate of content sharing*Polarization*Fake news
Fake news = 1
Demand for independent journalism = Misinformation
Revenue = Misinformation-Brand value
Brand value = 0.5*Revenue
Testing:
Policy Design and Evaluation:
Scenario Specification:
The historical data suggests that before the era of social media arrived, people used to rely on traditional news media systems such as articles, journals and newspapers. Polarization and confirmation bias was still taking place in the era between 1800-1820. But due to increase in digital media such as social media, audios, videos and television it increased and doubled the rate.
However, due to the expansion of social media which lacks the ability to fact check or the source of the information, has exposed people to the content which may lack truth and evidence. The driving force behind this is the decentralization of the power behind who can share news or content. Though this helps avoid the concentration of media outlets, it also makes it vulnerable to malicious users. Moreover, the reason behind the confirmation bias and polarization is the algorithms which are designed to feed content based on the user's search history, content shared with others, which creates a echo chamber of biased ideas among its users.
Policy Design
The purpose of the models created was to show how knowingly or unknowingly, this algorithm is being exploited by the political campaigns and how it only helps the social media companies to grow and gain their strong hold among users.
It can be seen from the models provided in the document that the social media platform becomes biased toward a political party that spends more on the ads on these platforms. Which results in misleading their users with biased news. Moreover, the anonymity on the internet allows its users to share or comment content which otherwise would not hold water in the real world. Also, social media has been one of the reasons for people isolating themselves from the outside world and spending hours on these platforms which may or may not depict the accurate picture of the real world. And this dopamine driven feedback loop increases the use of social media and ultimately its revenues. As it can be seen from the graph that there are nearly 3.02 billion users on social media which keeps on increasing and which results in polarization. As a result of this, the revenue of social media increases exponentially.
However, with recent political events on whistleblower from the ex employees of these companies, the demand of improving these platforms has been raised along with independent journalism. If social media such as facebook is forced to shift their focus from these revenue sources and and starts to show partisan news which are unbiased towards particular political parties, then people will find the news that are contrasting towards their opinions and it will eventually lead to depolarization and people will make less use of social media, hence decreasing the social media revenue and market share. Also, if people find the news fake, which are being shared on digital platforms then their like, view and share will decrease and eventually that would lead to decrease in their brand value and which leads to depolarization.
References
Levy, Ro'ee. 2021. "Social Media, News Consumption, and Polarization: Evidence from a Field Experiment." American Economic Review, 111 (3): 831-70.
DOI: 10.1257/aer.20191777
https://www.nature.com/articles/s41598-020-62085-w#citeas
https://www.nature.com/articles/srep40391 - Del Vicario, M. et al. Modeling confirmation bias and polarization. Sci. Rep. 7, 40391; doi: 10.1038/srep40391 (2017).
The role of (social) media in political polarization: a systematic review (tandfonline.com)
Tucker, J. A., Guess, A., Barber, P., Vaccari, C., Siegel, A., Sanovich, S., ... & Nyhan, B. (2018). Social media, political polarization, and political disinformation: A review of the scientific literature. Political polarization, and political disinformation: a review of the scientific literature (March 19, 2018).
Guerra, P. C., Meira Jr, W., Cardie, C., & Kleinberg, R. (2013, June). A measure of polarization on social media networks based on community boundaries. In Seventh international AAAI conference on weblogs and social media.
Bail, C. A., Argyle, L. P., Brown, T. W., Bumpus, J. P., Chen, H., Hunzaker, M. F., ... & Volfovsky, A. (2018). Exposure to opposing views on social media can increase political polarization. Proceedings of the National Academy of Sciences, 115(37), 9216-9221.
Goya-Tocchetto, D., Kay, A. C., Vuletich, H., Vonasch, A., & Payne, K. (2022). The partisan trade-off bias: When political polarization meets policy trade-offs. Journal of Experimental Social Psychology, 98, 104231.
Yang, H. C., & DeHart, J. L. (2016). Social media use and online political participation among college students during the US election 2012. Social Media+ Society, 2(1), 2056305115623802.
https://knowledge4policy.ec.europa.eu/visualisation/number-social-media-users-worldwide-2010-17-forecasts-2021_enData about spending on Social Media use and ads
https://www.statista.com/statistics/309592/online-political-ad-spend-usa/https://www.statista.com/study/40641/social-media-and-politics-in-the-united-states/https://www.statista.com/statistics/330302/sources-of-government-and-politics-news-usa/ (use as a balancing loop as the number of users who gets political news is around 1%)
https://www.statista.com/statistics/184541/typical-daily-online-activities-of-adult-internet-users-in-the-us/
https://www.sciencedirect.com/science/article/abs/pii/S0747563221000819https://www.tandfonline.com/doi/full/10.1080/01972243.2018.1497743 https://www.pewresearch.org/politics/2015/09/03/the-whys-and-hows-of-generations-research/
http://briswa.eu/the-model/causal-loop-diagram.htmlhttps://www.researchgate.net/figure/Information-seeking-causal-loop-diagram_fig1_259458221https://metasd.com/2020/10/systems-thinkers-social-media/https://www.statista.com/statistics/272014/global-social-networks-ranked-by-number-of-users/https://www.statista.com/statistics/1279138/united-states-source-facebook-news-content-views/
