Executive Summary

Full report

On this page

This report outlines a comprehensive research plan to identify and prioritize the most salient topics related to U.S. interest in rebellion and political unrest (search interest, scholarly research, media coverage, social signals, and public opinion) from 2018 through October 16, 2026. We will combine multiple data sources – Google Trends (search queries), academic databases (publication counts, SSRN, Google Scholar), news/media aggregators (GDELT, LexisNexis/Factiva), social media APIs (Twitter/X, CrowdTangle), and polling (Pew, NORC, Gallup, etc.) – to measure “interest” in various facets of political unrest. Our approach includes time-series analyses (e.g. query popularity over 2018–2026), event counts (news articles, protest incidents), social engagement metrics (shares, mentions, hashtags), and survey questions on grievances and violence.

We will identify the top 5 topics by combining indicators: search volume (Trends), news volume (LexisNexis counts), social media trends (tweet volumes, CrowdTangle engagements), and academic attention (publication counts or citations on related keywords). For each top topic, we provide a focused research question, key metrics, data sources, methodology, time windows, potential biases, and expected deliverables. We also present a ranked list of 10 self-contained research prompts (title + brief scope) that cover important angles of U.S. unrest and rebellion interest (e.g. economic causes, demographics, online mobilization, militia activity, protest escalation, policy response, misinformation, etc.). Throughout, we emphasize primary data and peer-reviewed work.

Key findings from preliminary review: U.S. search interest in terms like “civil war” or “insurrection” surged around national elections, indicating broad public curiosity (e.g. a 2024 Rasmussen poll found ~41% of likely voters expect a civil war in the next five years). Public opinion surveys (Pew, UC Davis) show large majorities perceive political violence as increasing and distrust institutions (85% say violence is rising; >80% say elected officials “don’t care”), even though only a small minority (<15%) report willingness to personally use violence. Social-science studies have explicitly linked online mobilization (e.g. activity on fringe social platforms) to offline unrest. These suggest “search interest in civil war”, “social media radicalization”, “support for militias and violence”, and “political trust/distrust” will emerge as top topics.

The deliverables will include tables and charts comparing key metrics across topics (e.g. trend graphs of query volume, news article counts, hashtag usage), a mermaid diagram linking causes and effects of unrest (e.g. how economic stress ➔ online grievance ➔ mobilization), and an executive summary highlighting the prioritized research agenda. We assume necessary data access (APIs for Trends, GDELT, Twitter, etc.) and list these as assumptions. The outline below details the research plan for each top topic, followed by the ranked list of 10 research prompts.

---

1. Search Interest in “Civil War” and Insurrection

(1) Research question: Are Americans increasingly searching for terms related to rebellion (e.g. “civil war,” “insurrection,” “uprising”)? How do search trends reflect spikes in concern or interest (e.g. around elections or crises)?

(2) Key metrics: Google Trends indexes (relative search volume) for queries like “civil war,” “second civil war,” “insurrection,” “revolution,” “overthrow government,” “citizen militia” in the US; search volume by state/region (heatmap); trends on social platforms (Twitter hashtags like #CivilWar, #Insurrection); counts of news articles (LexisNexis, GDELT) containing these terms; academic article counts (WoS/Scopus/Google Scholar) on “civil war USA 21st century,” etc. Public-opinion poll questions on civil war likelihood (e.g. “How likely do you think a second Civil War will occur?”).

(3) Data sources: Google Trends API (to query and download historical weekly/monthly search index for each keyword); GDELT Global Knowledge Graph for news mentions of key terms; LexisNexis or Factiva to count news/media articles by year containing these terms; Twitter/X API or CrowdTangle for social media volume/engagement on related hashtags and phrases; Pew Research Center or Rasmussen Reports for any polling questions on civil war likelihood; SSRN/Google Scholar/Web of Science for counts of academic papers on this topic. (We assume API access and keyword matching, with synonyms and boolean queries.)

(4) Methodology: Query Google Trends for each term (normalize to common baseline, e.g. index 0–100 relative to peak). Compare annual/quarterly trends (2018–2020 vs 2021–2023 vs 2024–Oct 2026) to see spikes (e.g. Nov 2020, Jan 2021, midterm elections). Use GDELT to retrieve weekly counts of news stories containing “civil war” or “insurrection” (with filters for U.S. context). On Twitter/CrowdTangle, track frequency of hashtag mentions (time series) and engagement (likes/shares). For polls, track question wording and year-to-year responses. We will cross-check Google Trends signals against known events (e.g. Jan 6, election periods) and against normal seasonal baselines. Charts: line graphs of search index over time for each term, comparative overlays (e.g. “civil war” vs “revolution”). Heatmaps: state-level interest (Trends interest by subregion). Tables: top terms by peak search volume, year-over-year growth.

(5) Time windows & baselines: We analyze three periods: 2018–2020 (pre-pandemic/election), 2021–2023 (post-2020 election period), and 2024–Oct 2026. Compare 2024–26 trends to the same months in 2019–21 as baseline. Key events (Nov 2020, Jan 2021, Jun 2024 primaries, Nov 2024 election) will be annotated. We will also consider smaller windows (monthly) around trigger events.

(6) Potential biases/limitations: Google Trends data is relative and anonymized (not raw counts); comparisons across unrelated terms require careful normalization (the index is 0–100 per query). Popular culture (e.g. movies titled “Civil War”) can confound queries. Social media data is subject to bots and platform moderation (Twitter bans certain accounts). News counts may reflect editorial bias (some terms appear more in sensational stories). Polls like Rasmussen may not be fully representative; question wording matters (e.g. asking likely vs justified). We must guard against overinterpreting short-lived spikes (e.g. viral memes).

(7) Deliverables:

• Table: Annual search interest index (or query hits) for each key term by year and region.
• Chart: Time-series graph of the Google Trends index for “civil war,” “insurrection,” etc., highlighting major peaks.
• Heatmap: Geographic breakdown of interest (e.g. search interest by state).
• Mermaid diagram: Causal map linking events ➔ spikes in search interest (e.g. Election fear → “Civil war” searches).
• Executive summary describing which queries show the largest growth and when.

(8) Effort & data needs: ~1–2 weeks to design queries, download Trends data (multiple terms, regions), verify API usage. Data needs: Google Trends API, GDELT API or CSV export, LexisNexis query, Twitter API credentials (or CrowdTangle with keyword filters). Poll data may require aggregating multiple surveys (Rasmussen, etc.). Effort includes cleaning and normalizing data (e.g. adjusting for Google sampling) and plotting charts.

---

2. Public Polling on Violence and Rebellion

(1) Research question: How many Americans express support for or willingness to engage in protest, insurrection, or political violence? Are these attitudes changing over time?

(2) Key metrics: Survey percentages of respondents endorsing political violence or rebellion (e.g. “violence is justified to achieve X goal,” “I would use force to defend my beliefs”); belief that a civil war is likely or necessary; support for extremist organizations (e.g. Proud Boys, militia groups); trust/distrust in democratic institutions (as a proxy grievance measure). Metrics include poll response rates by party/demographic, changes in support for violence year-to-year, and attitudes toward specific tactics.

(3) Data sources: Public-opinion polls and longitudinal surveys: UC Davis’s Center for Violence Prevention annual survey (2022–2025); Pew Research Center surveys on political violence (e.g. Oct 2025 found 85% say violence is rising); NORC’s General Social Survey or American National Election Studies for trust metrics; Gallup/NORC polls on protest; governmental surveys (Harris, YouGov, etc.) if available via aggregators. We will also query ACLED (which includes public protest events, though mostly non-violent) and Count Love (for US protests) for event counts as contextual background. Note poll question wording (willingness vs justification vs likelihood).

(4) Methodology: Compile a timeline of relevant survey questions (2018–2026) on support for violence or rebellion. Standardize metrics (e.g. percent “strongly/very willing” to use violence). Analyze changes by demographic slice (age, ideology, MAGA vs others). Compare attitudes before/after key events (e.g. Jan 6, election). Use cross-tabulation to identify groups with higher endorsement (e.g. [29] shows MAGA Republicans much more likely than Democrats to say violence is justified). Charts: multi-line chart of percent supporting violence (by year). Tables: poll results side-by-side, broken by subgroup. We will also correlate “support for overthrow” poll items (e.g. Wintemute 2025 found ~10% willing to use violence to support/oppose gov’t).

(5) Time windows & baselines: We will compare pre-2020 (baseline) to post-2020 trends, splitting into 2018–2020, 2021–2023, and 2024–2026. Key junctures: 2020 election, Jan 6, 2022 midterms, 2024 primaries/election. Baseline polls from early 2000s (Gallup, Pew older data on trust) can provide context (e.g. trust in government was higher historically).

(6) Potential biases/limitations: Self-reported willingness to use violence may understate true propensity (social desirability bias). Poll samples may miss extremist segments. Poll wording (e.g. “would you consider it justified” vs “would you do it yourself”) yields different rates. Many polls oversample likely voters, skewing older. Survey timing (in wake of attacks or political events) can cause spikes. “Likely voter” polls (Rasmussen) may differ from general adult population.

(7) Deliverables:

• Table: Selected poll questions and year, showing percent support for violence or rebellion by party and age.
• Chart: Trends in key attitudes (e.g. “violence justified” and “willing to use force”) from 2018–2026.
• Chart: Party breakdown (stacked bars) for one key item (e.g. “violence is justified in some cases”) in most recent survey.
• Mermaid diagram: Flow of grievances → willingness to rebel (e.g. distrust in election fairness → support for violence).
• Summary of trend findings (e.g. “surveys show no large increase in personal willingness to commit violence, consistent with UC Davis finding of ‘relative stability’, but a small minority endorses force under some scenarios”).

(8) Effort & data needs: ~2 weeks to gather poll datasets (from Pew, UC Davis, Gallup, etc.), clean and align question wording, and run trend analyses. Data: historical polls (many available online), knowledge of weighting. Will require coding poll responses for comparability.

---

3. Social Media and Online Mobilization

(1) Research question: How does online extremism and misinformation relate to offline unrest? Are spikes in activity on fringe platforms (Parler, Gab, 4chan, encrypted channels) associated with protest events?

(2) Key metrics: Volume of posts or engagement on “hard-right” or extreme-content platforms (e.g. Parler, Gab, Twitter alt-tech), mentions of keywords (e.g. “Stop the Steal,” conspiracy hashtags), follower growth of extremist influencers; correlation of this activity with counts of offline incidents (ACLED protest/violence events). Metrics include monthly user activity (posts/day), hashtag frequency, sentiment analysis on extremist hashtags, and follower counts. Also citations of conspiracy terms (“QAnon,” “false flag”) as proxies for mobilization interest.

(3) Data sources: CrowdCounting Consortium or ACLED for offline protest event counts by date and region (as used in Karell et al.); Parler/Gab archives or Hornetfeed data for 2020–2022 (some are public datasets); Twitter API for keywords (“militia”, “civil war”), including Twitter’s dataset of known extremist accounts; CrowdTangle for Facebook/Instagram mentions of relevant pages. Potential use of Telegram channel data (via APIs or scraping) if available. Scholarly data: the Yale study and other sociological analyses.

(4) Methodology: Conduct Granger-type time-series analysis: e.g. regress monthly counts of civil-unrest events (from ACLED/CrowdCounting) on prior months’ counts of extremist social media posts (following Karell’s approach). Topic modeling on collected posts to identify emerging calls to action. Sentiment or rhetoric analysis on mainstream vs fringe platforms. Charts: overlay graphs of social media activity vs protest events (with lags). Network graph: connections among extremist accounts. Example: “10% rise in Parler posts predicted 0.04% rise in far-right unrest events next month”.

(5) Time windows & baselines: Focus especially on Jan 2020–Dec 2022 (when Parler data is available) and 2023–Oct 2026 for Telegram/Twitter alt-tech. Baseline: earlier social-media activity (e.g. 2017–2019) if data exists. Key comparison: pre- vs post-platform bans (Twitter fact-checking, Parler shutdown).

(6) Potential biases/limitations: Extremist content often moves to encrypted or private channels (harder to measure). Data from closed platforms may be incomplete. Engagement counts can be inflated by bots (especially on fringe sites). Causal inference is tricky: users may become more active after events rather than before. Privacy and ethical limits restrict data collection (need to avoid doxxing or direct quoting illegal content). Sampling bias: not all users opt-in to CrowdTangle.

(7) Deliverables:

• Table: Monthly counts of extremist posts (by platform) and protests (ACLED) side by side.
• Chart: Time series of hard-right social media engagement vs number of reported far-right protest events (with a lag line to illustrate lead-lag relationship).
• Mermaid: Network diagram of information flow (e.g. Echo Chambers on fringe platforms ➔ Norm shift ➔ Real-world action) based on social influence theory.
• Text summary linking our findings to the Yale study: “In line with Karell et al. (2023), we would expect even small surges in extreme online chatter to precede limited increases in unrest.”

(8) Effort & data needs: ~2–3 weeks. Data needs include API access or archives for Parler/Gab, Twitter historical counts, CrowdTangle login. Ability to use GDELT or ACLED export of protests. Will need network-analysis tools (Gephi or Python libraries) and time-series stats (Granger tests).

---

4. Militia and Extremist Organizations

(1) Research question: What are the trends in militancy and paramilitary movements? Is public membership or support for militia groups growing? How visible are these organizations in the media?

(2) Key metrics: Number of self-identified militia members or groups (as reported by SPLC/ADL); volume of news/media mentions of key groups (Oath Keepers, Proud Boys, Three Percenters); sales of military-style firearms (proxy); social media group membership counts; search volume for group names. Also poll support: percent of respondents approving militia groups (e.g. 14% support “private militias” enforcing policies). Online engagement: subreddit subscriber counts.

(3) Data sources: Southern Poverty Law Center (SPLC) and Anti-Defamation League (ADL) annual reports on extremist groups (they track number of militia groups and members); the FBI/DHS annual domestic terrorism reports (2022, 2023); search interest (Trends) for militia names; LexisNexis news search counts for group names; Twitter follower counts for these groups’ accounts. Pew/NORC polls if any on support for militia. (NORC did a recent survey on “militia and Proud Boys” attitudes, though access may be limited.)

(4) Methodology: Compile longitudinal data (2018–2026) on militia group metrics: e.g. SPLC historical counts of active groups; ACLED data on armed confrontations involving militias; measure spikes after high-profile events (e.g. Jan 6, election year). Use Google Trends to chart interest in “militia” vs “gun rights”. News count: number of stories per year mentioning “militia” or specific groups. Table: membership trends (from ADL/SPLC). Chart: news volume by year. We may overlay major political milestones.

(5) Time windows & baselines: Compare pre-2016 data (historic baseline) to post-2016 growth (SPLC notes uptick since 2016). Focus on 2018–2026 with pre-2020 vs post-2020 comparisons. Key baseline: 2000s era militia data. Policy events: after Jan6 arrests, 2022 midterms, 2024 debates.

(6) Potential biases/limitations: Militia membership is secretive; SPLC/ADL estimates may undercount or lag actual strength. News coverage can spike after single events (media effect). Google Trends for “militia” may capture hunting references. The term “militia” can be politically loaded and interpreted differently by Democrats vs Republicans.

(7) Deliverables:

• Table: Yearly count of known militia organizations (SPLC/ADL data) and FBI domestic terrorism incidents by ideology.
• Chart: News article frequency for “militia group” vs “gun owner” (baseline term) per year.
• Chart: Google search index for “militia”, “Oath Keepers”, etc., 2018–2026.
• Mermaid: Link between political events and militia mobilization (e.g. Election disputes ➔ militia organizing ➔ armed protests).
• Write-up summarizing whether raw data show sustained growth or episodic surges (for instance, a CSIS analysis noted a sharp drop in recent right-wing attacks in early 2025, which might reduce militia activism).

(8) Effort & data needs: ~1–2 weeks. Data: SPLC/ADL annual reports (publicly released), LexisNexis news database search (requires login), Google Trends. Possibly NRA membership numbers (for context). Analysis mostly descriptive (counts over time).

---

5. Political Distrust, Grievances, and Social Polarization

(1) Research question: How do measures of political discontent (distrust in government, belief in corruption or unfairness) relate to interest in rebellion? Are increasing negative attitudes toward institutions correlated with unrest interest?

(2) Key metrics: Poll results on trust in government, belief that elections were stolen, feelings of alienation (Pew trust indices, e.g. only 16% trust government); social media sentiment analysis of terms like “rigged,” “corrupt,” or “drain the swamp”; “extremism” concern levels (Pew: ~52% see both left- and right-wing extremism as major problems). Also number of mentions of “decline of democracy” or “anti-establishment” in news. Economic stress proxies: income inequality, unemployment by region (from Bureau of Labor).

(3) Data sources: Pew Research surveys on trust and cynicism (2023–2026); Gallup questions on trust in institutions; General Social Survey trends on confidence in Congress/Government; news archives for “rigged election” references; NORC/PRRI data on populist attitudes; World Values Survey if recent US data exists. Social media: tweets with words like “corrupt politicians,” “tyranny,” to quantify grievance.

(4) Methodology: Track year-by-year trust indices and compare to spikes in unrest metrics (from above). Geospatial: map trust levels by state/region (if polling allows) and compare to protest hot spots (ACLED data). Sentiment analysis: e.g. proportion of tweets with negative sentiment about government (using a lexicon approach). Table: Pew trust scores over time (2018 vs 2023). Chart: overlay of trust level and some unrest metric (e.g. # protests or violence events). Topic modeling: identify prevalent grievances (immigration, economy) in news/social text related to unrest.

(5) Time windows & baselines: Compare “distrust” trends over the same periods (2018–20, 2021–23, 2024–26). Pew’s 2023 surveys and Spring 2025 discontent reports provide baselines. Historic: in 1970s, US trust was much higher; a baseline fact (“trust ~40% in 1970s vs ~20% now”) can be cited for context.

(6) Potential biases/limitations: Self-reported trust is subjective; question wording (feelings-based) influences results. Regional polling may be sparse. Online sentiment could be dominated by activists and bots, not representative. News media focus on corruption could inflate apparent public concern. Causality is murky: distrust might motivate protest, or frequent exposure to protest might increase distrust.

(7) Deliverables:

• Chart: Timeline of institutional trust (Pew % who trust government) vs timeline of unrest indicators (e.g. protest frequency).
• Heatmap: State-level trust scores (if data available) compared to protest per capita.
• Table: Key survey findings (e.g. “>80% say officials don’t care about people like them”; “85% think violence is rising”).
• Mermaid diagram: System map linking distrust (e.g. low trust → grievance narratives → mobilization).
• Narrative on whether rising distrust has coincided with greater unrest interest (e.g. Pew shows no consensus that distrust has eased, which could sustain interest in drastic solutions).

(8) Effort & data needs: ~1–2 weeks. Data: Pew and Gallup historical datasets, SGP Polling, news sentiment via GDELT (which can export frequency of terms like “lock them up,” “civil war”), Census data for economic covariates. Analysis involves survey trend analysis and GIS mapping.

---

Ranked List of 10 Research Prompts (U.S. Rebellion Interest Facets)

Below is a prioritized list of 10 self-contained research prompts (title + 2-sentence scope) covering critical facets of U.S. interest in rebellion and unrest. These can each form the basis of independent reports:

1. “American Fury: Economic Grievances and Unrest.” Investigate how economic inequality, job loss, and inflation have fueled interest in protest and fringe political movements. Analyze correlations between regional economic stress indicators and rises in protest events or search interest in rebellion.

1. “Youth Radicalization and Political Disillusionment.” Examine whether younger Americans are more likely to embrace insurgent ideas. Study generational differences in online search queries and social media discourse about overthrowing government or violent protest.

1. “Online Echo Chambers and Mobilization.” Assess how social media algorithms and fringe platforms amplify extremist content. Measure whether spikes in online conspiratorial chatter (e.g. QAnon, election fraud claims) precede mass mobilizations or violent incidents.

1. “Militia Movements and Paramilitary Organizing.” Analyze trends in domestic armed groups (e.g. Oath Keepers, Three Percenters). Include counts of chapters and membership (from SPLC/ADL), levels of training exercises, and any official crackdowns to gauge momentum.

1. “Escalation of Protest Tactics.” Trace how U.S. protest movements have shifted tactics over 2018–2026 (from peaceful marches to property damage to armed confrontation). Use ACLED and media reports to classify event severity and chart any escalation.

1. “Government Responses and Policy Repercussions.” Study how law enforcement and legislative actions (e.g. anti-riot bills, social media moderation, pardons) affect public sentiment. Do crackdowns reduce search interest in rebellion or fuel conspiracy narratives?

1. “Demographics of Discontent.” Identify which demographic groups (age, race, education, geography, party) show higher interest in rebellion-related content. Use polls, voter surveys, and social-media geo-tags to map the sociopolitical profile of the disaffected.

1. “Misinformation and Conspiracy Propagation.” Analyze the role of false narratives (e.g. “stolen election,” “Great Replacement”) in driving unrest interest. Use fact-check databases and social-media misinformation trackers to see when and where these narratives surge.

1. “Election Denial and Democratic Stability.” Examine how beliefs about election fraud correlate with insurrectionist sentiments. Assess polls on election confidence and count incidents of violence or civil disobedience around election cycles.

1. “Comparative International Perspective.” Compare U.S. rebellion interest with other democracies (e.g. France Yellow Vests, Brazil protests). Determine whether the U.S. is unique or part of a broader pattern of democratic discontent in this period, using comparative polls and event data.

---

All proposed analyses will rely primarily on official data and peer-reviewed research. We will prepare summary tables comparing the top topics on each metric (e.g. a table of Google Trends’ relative search volumes, news-article counts, social mentions, and academic papers for each topic). Charts will include multi-year time series of key indicators, geographic heatmaps (e.g. protest intensity by state), and mermaid diagrams illustrating causal pathways. An executive summary will highlight the 3–5 strongest signals of public interest (e.g. unprecedented Google Trends peaks for “civil war”, majority perception of rising political violence, growth in left-wing vs right-wing extremist events). A clear research plan and deliverables checklist will be provided for each topic.