Earthquake conspiracy discussion on Twitter

The study examines how discourse about the HAARP conspiracy theory evolves on Twitter and identifies catalysts for heightened attention, particularly around major seismic events. Conspiracy theories tend to proliferate after unexpected, high-impact events as people seek explanations and assign blame. Social media facilitates rapid dissemination and community formation around such narratives, potentially amplifying misinformation. The 7.8 magnitude earthquake on February 6, 2023 in Turkey and Syria serves as a focal event. The research question is: how does HAARP conspiracy discourse change over time on Twitter, and what triggers surges in attention? Understanding these dynamics can inform strategies to mitigate misinformation and illuminate the role of digital communication in belief formation during crises.

Background highlights how social media platforms (e.g., Twitter, Facebook, YouTube) enable rapid spread of conspiracy theories through engagement-driven algorithms and like-minded communities. Prior work links major crises (e.g., COVID-19) to surges in conspiratorial narratives, with consequences such as vaccine hesitancy and increased distrust in authorities. Earthquakes are frequently targeted by conspiratorial claims attributing natural events to secret technologies or actors; HAARP is commonly alleged to control weather or cause disasters. The Turkey–Syria earthquake saw a large spike in earthquake-related tweets (about 1.5 million containing the word earthquake on February 6, 2023), alongside widespread HAARP-related claims. This context situates HAARP within broader patterns of online conspiracy proliferation after disasters.

Design: Case study approach focusing on HAARP discourse and the Turkey–Syria earthquake. Data sources: Twitter API v2 for Academic Research; USGS Significant Earthquakes Archive; Google Trends for auxiliary validation of public interest. Time windows: - Frequency analysis: January 1, 2022–March 4, 2023, capturing nearly all available tweets containing the keyword “HAARP.” - Sentiment analysis: January 1, 2023–February 28, 2023 on English-language tweets only. Data scope: 1,041,633 HAARP tweets total across the study period. Analytic steps: - Extract daily tweet frequencies for “HAARP.” - Identify and qualitatively examine peak days to infer potential triggers (e.g., major earthquakes, viral videos, campaign announcements). - Correlate daily HAARP tweet counts with the maximum earthquake magnitude per day (USGS) over Jan 2022–Mar 2023. - Perform sentiment analysis (Microsoft Azure Text Analytics) on English tweets Jan–Feb 2023; sentiment scores range 0–1 (higher is more positive; 0.45–0.60 neutral). - Compute correlations among tweet volume, earthquake magnitude, mean sentiment, and sentiment variability; compare pre-/post-earthquake sentiment. - Create word clouds of top terms before (Jan 1–Feb 5, 2023) and after (Feb 6–28, 2023) the earthquake to characterize topics. Notes: Frequency analysis includes all languages; sentiment analysis limited to English. An AI-generated illustration (Midjourney) was created to accompany the HAARP conspiracy depiction (not part of analysis).

• Volume and peaks: Total HAARP tweets in the study period: 1,041,633. Monthly totals show a dramatic spike in February 2023 (678,641 tweets). Daily statistics (Jan 1, 2022–Mar 4, 2023): mean 2,434 tweets/day, max 150,530 (Feb 6, 2023). - Triggers: Peaks in HAARP discussion align with major earthquakes: Feb 6, 2023 (Turkey–Syria, M7.8; >150,000 HAARP tweets); Nov 23, 2022 (Turkey, M6.1; ~11,700 tweets); Sep 9, 2022 (Mexico, M7.7; >6,100 tweets). Other peaks relate to heavy rainfall in Brazil (Nov 2022), a viral video on China’s weather-modification system (July 2022), and a HAARP–NASA campaign (Dec 2022). - Correlation with earthquakes: Across Jan 2022–Mar 2023, the correlation between daily HAARP tweet count and maximum earthquake magnitude is positive and significant (r ≈ 0.148, p<0.001), indicating more intense earthquakes are associated with more HAARP discussion. - Sentiment–volume relationship (Jan–Feb 2023): Positive and significant correlation between number of HAARP tweets and mean sentiment (r ≈ 0.2247, p<0.1), suggesting discourse may reinforce beliefs and become more positively valenced as volume grows. Mean sentiment and its standard deviation are negatively correlated (r ≈ −0.2975, p<0.05), implying sentiment dispersion narrows as average sentiment increases. - Temporal sentiment shift: Mean sentiment was lower than the Jan–Feb average before the Feb 6 earthquake and increased afterward. - Topics: Before Feb 6, frequent terms centered on weather manipulation, climate control, chemtrails, geoengineering, and government involvement. Afterward, discussion included alleged HAARP roles in seismic events and geophysical warfare, with mixed skepticism and endorsement. - External validation: Google Trends searches for HAARP peaked following the Feb 2023 earthquake, aligning with Twitter activity; Turkey led search interest, followed by Albania, Kosovo, Lebanon, Cyprus, Bangladesh, Romania, North Macedonia, Pakistan, and Azerbaijan.

Findings indicate that major earthquakes act as catalysts for HAARP conspiracy discourse on Twitter. The positive association between tweet volume and average sentiment suggests belief reinforcement dynamics: as users encounter more like-minded content, positivity toward the narrative grows, potentially aided by echo chamber effects. While correlations cannot establish causality, the temporal alignment and known event sequence (earthquakes preceding discourse surges) support the interpretation that seismic events trigger HAARP-related attention rather than the reverse. Cross-platform indicators (Google Trends) corroborate heightened public interest following the Turkey–Syria earthquake. The results contribute to understanding how crisis events and platform dynamics can amplify conspiratorial narratives, influencing public perception and potentially hindering effective crisis communication.

The study shows that HAARP conspiracy discussions on Twitter intensify following major earthquakes, with the February 6, 2023 Turkey–Syria event producing the largest surge. Tweet volume correlates positively with mean sentiment, indicating discourse may reinforce beliefs and become more positively valenced as attention grows. These insights underscore the role of social media in amplifying conspiracy narratives during crises and highlight the need for timely, accurate information and critical thinking interventions. Future research could apply multivariate or causal inference approaches, incorporate multiple platforms and languages, integrate geographic inference, and improve sentiment modeling tailored to short, context-rich social media texts to better capture sarcasm and domain-specific nuances.

• Single-platform and language scope: Sentiment analysis limited to English tweets on Twitter; frequency analysis used all languages but still single platform. - Geographic uncertainty: Lack of precise location data; potential mitigation via alternative metadata or external sources (e.g., Google Trends), but residual noise remains. - Sentiment model constraints: Azure model trained primarily on product/service reviews may underperform with domain-specific language, sarcasm, dialects, or context-dependence in tweets. - Correlation vs causation: Pearson correlations cannot establish causal mechanisms; multivariate or causal designs would offer deeper inference. - Data artifacts and formatting issues: Inherent uncertainties in social media data (bots, misinformation, varying engagement) and possible measurement noise.