A Plurality Atlas

What is polarization?

The process by which people in a society split into opposing camps and come to dislike and distrust the other side more and more. The worry that social media has deepened division is one of the starting points of the book Plurality, but researchers dispute its role, and the book says so too.

We often hear that society today is divided. But what exactly is divided, whether things really have become more divided, and whether social media is to blame are surprisingly unclear. Researchers have taken up these questions in many surveys and experiments.

In short

Polarization is the process by which people in a society split into opposing camps and come to dislike and distrust the other side more and more.

Research separates two things. One is people's positions on issues such as policies drifting apart toward two poles. The other, separate from differences in position, is coming to dislike and distrust people who support the other party; this is called affective polarization [1].

The book Plurality takes as one of its starting points the worry that technologies such as social media have deepened division. But the book itself also says that the role of social media is "a topic of significant academic debate" [2].

For example

The United States. In 1960, about 5% of Republican and Democratic supporters said they would be displeased if their son or daughter married a supporter of the other party. By 2010, nearly 50% of Republicans and over 30% of Democrats said they would feel somewhat or very unhappy about it (according to a 2012 study by Iyengar and others, as cited by Boxell and others) [3]. A study comparing about four decades in twelve countries of the Organisation for Economic Co-operation and Development (OECD) found, in its baseline estimates, that affective polarization grew the most in the US. Six countries, including Japan, Britain and Germany, by contrast showed a downward trend (the decline was statistically significant only for Sweden and Germany) [4].

Taiwan. In a 2021 survey of 17 places by the US-based Pew Research Center, Taiwan had the smallest share of people saying there were conflicts between people who practice different religions, at 12%, and was also on the low side for perceived conflict between urban and rural residents and between racial and ethnic groups. Yet 69% said there were conflicts between people who support different political parties, which put Taiwan among the three most politically divided places surveyed [5].

A study using Taiwanese election surveys reports that the conflict between the sides of Taiwan's two main parties, the Kuomintang (KMT) and the Democratic Progressive Party (DPP), is not a split of positions into two poles on unification with China versus independence, but hostility between partisans. That hostility rose after the DPP first took power in 2000 (a change of ruling party) and eased from 2008 onward (an analysis of 1996–2012) [6]. Another study reports that voters who strongly dislike the opposing party increased sharply from 2012, reaching 40% of KMT and DPP supporters in 2020 [7].

Japan. Japan's 2019 White Paper on Information and Communications, from the Ministry of Internal Affairs and Communications, says that although there is research showing polarization advanced in the US over the past ten years, there are few empirical studies in Japan, and it cannot be concluded that Japan is polarized [8].

Germany. Steffen Mau and colleagues, the authors of Trigger Points, a book on inequality and political polarization (originally published in German in 2023), question the image of polarization in which society is changing from a one-humped camel (a dromedary) into a two-humped one (a camel). Summing up previous research, they say that in Germany, over the long term, there is little evidence that people's views have fundamentally split into two along a new line of conflict created by globalization, and that affective polarization there is below average in international comparison [9].

In more detail

Two kinds of polarization

According to a review of the political science research by Shanto Iyengar and colleagues, polarization used to be seen mainly in terms of issues. In the US in recent years, however, a new kind of division has emerged: ordinary people regard supporters of the other party as hypocritical, selfish and closed-minded, and avoid socializing across party lines [1]. The authors trace its roots to partisanship as a social identity — a sense of belonging to the group on one party's side [1].

Separating the two changes the picture. The Taiwanese studies above also examine polarization on issues and hostility between partisans separately [6][7].

What the book says

Chapter 2-0 of Plurality (English edition) describes technologies such as social media as being seen to break down the social fabric and heighten polarization, and calls such threats "anti-social" [2]. It then says that "polarization and the influence of extremist parties has been steadily rising in many developed democracies" [2].

But the book does not make a firm claim. It puts it this way: "While the role of the internet-based social media landscape is a topic of significant academic debate, recent surveys suggest that these tools have fallen far short of their promise of strengthening social and political bonds across differences and may well have contributed to the secular [long-term] rise in polarization since 2000, especially in the US" [2].

Chapter 5-4 says that "one of the most common concerns about social media has been its tendency to entrench existing social divisions, creating 'echo chambers'" [10]. According to the book, the algorithms that select and order the posts each user sees on social media (the news feed) pick content likely to maximize engagement and prioritize like-minded content. It goes on: "Despite mixed findings on whether these algorithms truly exacerbate political polarization and hamper deliberations, it is natural to ask how these systems might be redesigned with the opposite intention of 'bridging' the crowd" [10].

What the book says about Taiwan

Chapter 2-2 says that a variety of studies "have found that Taiwan is one of the least socially, ethnically and religiously polarized developed countries in the world, though some have found a slight upward trend in political polarization since the Sunflower movement" [11]. (The Sunflower Movement was a 2014 movement in which Taiwanese students and citizens, protesting that a trade agreement with China was being pushed through without proper review, occupied the Legislative Yuan, Taiwan's national legislature, for more than three weeks.) It adds that this is especially true of affective polarization, "with Taiwan consistently among the 5 least affectively polarized countries" [11].

What the international comparative paper the book cites as evidence says about Taiwan in its main text goes only this far: Taiwan "is among the least polarized nations in terms of feelings toward party leaders" [12].

Tang's view

Audrey Tang (唐鳳), who served as Taiwan's first Minister of Digital Affairs, said in a 2025 interview: "If you just look at the anti-social corner of social media, you will think that they are really polarized, but that's an illusion. Actually, most people agree with most of their neighbors, most of the time, on most of the things." [13] Tang also said that the Sunflower Movement ten years earlier could have turned violent, or dissipated, had depolarization technologies through bridging systems (tools that bring to the surface what people who usually disagree all support) not been deployed [13].

In a 2026 interview with the Swiss German-language newspaper Neue Zürcher Zeitung, Tang answered that polarization is energy. People fight over the most divisive points and overlook what they have in common; because social networks capture attention through outrage, a surprising common ground stays out of sight [14].

What is debated

Is social media the cause?

The empirical findings are split. Most of the research is from the US [15].

Studies that find harm. A 2022 systematic review of 496 studies found that associations with digital media that are likely to be harmful to democracy — such as declining trust, rising populism and growing polarization — were more pronounced in established democracies [15]. On polarization, it concluded that most studies show harmful associations and that the studies examining causes largely support them. But it also notes contrary findings, such as a two-month experiment in which exposure to news from the opposing side on Facebook reduced affective polarization [15].

Studies that find no visible effect. During the 2020 US presidential election, researchers ran an experiment together with Meta, which operates Facebook and Instagram. When the feeds of consenting users were switched from the default algorithmic ordering to newest-first (reverse-chronological) order, the time they spent on the platforms fell substantially and what they saw changed, but there was no significant change in levels of polarization over the three-month study period [16]. In an experiment with 23,377 Facebook users that reduced their exposure to like-minded sources by about one-third, there were also no measurable effects on eight preregistered measures, including affective polarization [17].

From the same period, however, there is also a study of aggregated data on 208 million US Facebook users. It reports that the degree to which conservatives and liberals were exposed to different political news was high, and grew stronger at each step: from all the news that could have appeared in their feeds, to the news actually shown after algorithmic curation, to the news users engaged with [18]. What this study examined was how separated the news people encountered was, not how polarized their views were.

Who became more polarized? One US study found that polarization grew most among the older groups that use the internet and social media least, and that, estimated with a model at the age-group level, the internet explains only a small share of the recent growth [3]. The twelve-country comparison above also found no positive association between the spread of the internet and trends in affective polarization. The authors themselves, however, write that comparing countries is difficult and that firm conclusions about causes are hard to reach [4].

In Japan. A survey that asked the same people twice (2017–2018) found that people who started using Facebook or Twitter showed no tendency to become politically more extreme, and that evidence of the internet driving polarization is scant [19]. The ministry's white paper also presents research suggesting that internet use may increase both the number of people with xenophobic views and the number opposed to them, and concludes that whether the internet itself drives polarization remains open to debate [8].

Plurality too uses phrases such as "a topic of significant academic debate", "may well have contributed" and "mixed findings", and does not claim outright that social media is the cause [2][10].

Does exposure to opposing views soften it?

In one experiment, US Democrats and Republicans who visit Twitter at least three times a week were paid to follow, for one month, an automated account (a bot) that passed on posts from politicians, opinion leaders and others on the opposing side (carried out in 2017, published in 2018). Republicans who saw liberal posts became substantially more conservative. Democrats who saw conservative posts became slightly more liberal, but this effect was not statistically significant [20]. The authors themselves note that the study has important limitations [20].

Is Taiwan really resistant to polarization?

As seen under "For example", the Pew survey the book cites matches the book's description where social and religious conflict are concerned, but for the share of people who perceive conflict between parties, Taiwan is among the top three places [5][11]. Within Taiwan too, there is research finding that strong antipathy toward the opposing party has spread since 2012 [7].

This part of the book is worded differently in different editions. The Japanese edition renders it as 最も政治的、社会的、宗教的に分断の低い (least politically, socially and religiously divided), putting "politically" where the English edition has "ethnically" [21]. The Traditional Mandarin edition adds, among other things, that political polarization has since eased, and that the dramatic behavior of political elites is not fully reflected in society [22].

Is the image of a divided society exaggerated?

For Germany, Mau and colleagues argue that the view of society as two camps is exaggerated, and that what happens instead is that conflict flares up suddenly over particular issues, which they call trigger points [9]. They also question whether analyses of the US can simply be applied to Germany [9].

Criticism of bridging designs themselves

Designs meant to cross divides are criticized too. Community Notes on X (formerly Twitter) lets users attach explanatory notes to posts that might mislead, and shows only the notes that people who usually disagree in their ratings both rate "helpful". One study argues that this system, precisely because it captures each country's main axis of division well, by design ends up showing fewer notes on highly polarizing content [23]. The details are in the Bridging and Community Notes articles.

Where it fits in Plurality

Plurality treats division as one face of the gulf between technology and democracy that it describes in chapter 2-0, "Information Technology and Democracy: A Widening Gulf" (in the Japanese edition, ITと民主主義 拡大する溝) [2]. It then presents augmented deliberation (using technology to extend people's ability to talk and think together at scale), Community Notes and Polis (a discussion tool in which people agree or disagree with short statements, bringing to the surface the statements that win support across opinion groups) in chapter 5-4 as designs that aim in the opposite direction [10]. The Japanese edition's title is PLURALITY 対立を創造に変える、協働テクノロジーと民主主義の未来 (Plurality: turning conflict into creation — the future of collaborative technology and democracy) [24].

Chapter 2-2 presents Taiwan as an example noted for its resistance to polarization [11]. As seen above, the picture differs depending on whether one looks at social conflict or at conflict between parties [5].

Chapter 5-4 cites a study to say that Community Notes "has been shown to encourage the exploration of diverse political information, compared to the previous methods of moderating misinformation" [10]. According to that study, posters on Twitter (X) who were tagged by individuals as spreading misinformation retreated into information bubbles. With collective tagging, reviewed by other fact-checkers as in Community Notes, that tendency was weaker [25]. What the study measured was not polarization itself but the diversity of the information people engaged with [25].

Related articles: - Bridging — a design that values support from across divides - Community Notes — a system that shows only the notes accepted by people who usually disagree - Prosocial media — a vision of extending bridging to the design of social media as a whole - Augmented Deliberation — the chapter where the book discusses alternatives to division - Taiwan's digital democracy — Taiwan, which the book presents as an example of resistance to polarization - The Sunflower Movement — the movement Tang cites as an example of depolarization technology

Further reading

  1. Plurality, chapter 2-0 "Information Technology and Democracy: A Widening Gulf" (English, book) — https://www.plurality.net/v/chapters/2-0/eng/
  2. Lorenz-Spreen and others, "A systematic review of worldwide causal and correlational evidence on digital media and democracy" (2022, English, paper) — https://doi.org/10.1038/s41562-022-01460-1
  3. Nyhan and others, "Like-minded sources on Facebook are prevalent but not polarizing" (2023, English, paper) — https://doi.org/10.1038/s41586-023-06297-w
  4. Iyengar and others, "The Origins and Consequences of Affective Polarization in the United States" (2019, English, paper) — https://doi.org/10.1146/annurev-polisci-051117-073034
  5. Ministry of Internal Affairs and Communications, White Paper on Information and Communications 2019, "インターネットの利用は世論を二極化するのか" (Does internet use polarize public opinion?) (Japanese, white paper) — https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r01/html/nd114200.html
  6. Tatsuo Tanaka (田中辰雄) and Satoshi Hamaya (浜屋敏), "ネットは社会を分断するのか" (Does the internet divide society?) (2018, Japanese, research report; archived copy on the Internet Archive) — http://web.archive.org/web/20210918143446/https://www.fujitsu.com/jp/group/fri/imagesgig5/no462.pdf

Sources

  1. Iyengar and others, "The Origins and Consequences of Affective Polarization in the United States" (2019) — https://doi.org/10.1146/annurev-polisci-051117-073034
  2. Plurality, 2-0 "Information Technology and Democracy: A Widening Gulf" — https://www.plurality.net/v/chapters/2-0/eng/ ; Japanese edition 「ITと民主主義 拡大する溝」 — https://www.plurality.net/v/chapters/2-0/jpn/ [CC0]
  3. Boxell, Gentzkow and Shapiro, "Greater Internet use is not associated with faster growth in political polarization among US demographic groups" (2017) — https://doi.org/10.1073/pnas.1706588114
  4. Boxell, Gentzkow and Shapiro, "Cross-Country Trends in Affective Polarization" (2024) — https://doi.org/10.1162/rest_a_01160
  5. Pew Research Center, "Diversity and Division in Advanced Economies" (2021-10-13) — https://www.pewresearch.org/global/2021/10/13/diversity-and-division-in-advanced-economies/
  6. Yi-ching Hsiao (蕭怡靖), "從政黨情感溫度計解析台灣民眾的政治極化" (Political Polarization in Taiwan: An Analysis on Mass Feeling Thermometer toward Political Parties) (Journal of Electoral Studies, 2014, in Chinese) — https://jestw.nccu.edu.tw/uploads/paper/tw/從政黨情感溫度計解析台灣民眾的政治極化.pdf
  7. Hung-i Lee and Yu-tzung Chang (李弘繹・張佑宗), "負面黨性與投票抉擇:2004-2020年臺灣總統選舉的分析" (Negative Partisanship and Voting Choice: The Case of Presidential Elections in Taiwan, 2004-2020) (Journal of Electoral Studies, 2022, in Chinese) — https://jestw.nccu.edu.tw/uploads/paper/tw/4pa62.pdf
  8. Ministry of Internal Affairs and Communications, White Paper on Information and Communications 2019, "インターネットの利用は世論を二極化するのか" (Does internet use polarize public opinion?) (in Japanese) — https://www.soumu.go.jp/johotsusintokei/whitepaper/ja/r01/html/nd114220.html
  9. Mau, Lux and Westheuser, Trigger Points, chapter 1 "Introduction" (2026, English edition) — https://doi.org/10.56687/9781529254051-004 [CC BY-NC-ND 4.0]
  10. Plurality, 5-4 "Augmented Deliberation" — https://www.plurality.net/v/chapters/5-4/eng/ [CC0]
  11. Plurality, 2-2 "The Life of a Digital Democracy" — https://www.plurality.net/v/chapters/2-2/eng/ [CC0]
  12. Reiljan and others, "Patterns of Affective Polarization toward Parties and Leaders across the Democratic World" (2024) — https://doi.org/10.1017/S0003055423000485 [CC BY 4.0]
  13. Audrey Tang, interview on PBS Firing Line (2025-10-11, transcript) — https://archive.tw/2025-10-11-interview-with-pbs-firing-line [CC0]
  14. Neue Zürcher Zeitung, "Taiwans Cyber-Botschafterin Audrey Tang über Demokratie im Zeitalter von KI" (Taiwan's cyber ambassador Audrey Tang on democracy in the age of AI) (2026-04-11, in German) — https://www.nzz.ch/technologie/polarisierung-setzt-energie-frei-aber-man-darf-sie-nicht-in-extreme-lenken-sagt-taiwans-cyber-sonderbotschafterin-audrey-tang-ld.1926179
  15. Lorenz-Spreen and others, "A systematic review of worldwide causal and correlational evidence on digital media and democracy" (2022) — https://doi.org/10.1038/s41562-022-01460-1 [CC BY 4.0]
  16. Guess and others, "How do social media feed algorithms affect attitudes and behavior in an election campaign?" (2023) — https://doi.org/10.1126/science.abp9364
  17. Nyhan and others, "Like-minded sources on Facebook are prevalent but not polarizing" (2023) — https://doi.org/10.1038/s41586-023-06297-w [CC BY 4.0]
  18. González-Bailón and others, "Asymmetric ideological segregation in exposure to political news on Facebook" (2023) — https://doi.org/10.1126/science.ade7138
  19. Tatsuo Tanaka (田中辰雄) and Satoshi Hamaya (浜屋敏), "ネットは社会を分断するのか -パネルデータからの考察-" (Does the internet divide society? An analysis of panel data) (Fujitsu Research Institute research report No. 462, 2018, in Japanese; archived copy on the Internet Archive) — http://web.archive.org/web/20210918143446/https://www.fujitsu.com/jp/group/fri/imagesgig5/no462.pdf
  20. Bail and others, "Exposure to opposing views on social media can increase political polarization" (2018) — https://doi.org/10.1073/pnas.1804840115 [CC BY-NC-ND 4.0]
  21. Plurality, Japanese edition, 2-2 "デジタル民主主義の日常" (The Life of a Digital Democracy) — https://www.plurality.net/v/chapters/2-2/jpn/ [CC0]
  22. Plurality, Traditional Mandarin edition, 2-2 "數位民主的日常" (The Life of a Digital Democracy) — https://www.plurality.net/v/chapters/2-2/zh-tw/ [CC0]
  23. Bouchaud and Ramaciotti, "Community Notes undermoderate polarizing content by design creating risks in electoral processes" (2025) — https://arxiv.org/abs/2506.15168 [CC BY-NC-ND 4.0]
  24. Ken Suzuki, commentary on the Japanese edition of Plurality (in Japanese) — https://note.com/digitaldemocracy/n/nb8b59a6b775d
  25. Kim and others, "Differential impact from individual versus collective misinformation tagging on the diversity of Twitter (X) information engagement and mobility" (first version 2023; published in Nature Communications in 2025) — https://arxiv.org/abs/2311.11282 [CC BY 4.0]

We welcome error reports and questions ("What is…?"). Submissions become public issues.

Report an error in this article ・ Ask a question