Visualization Tools

Viral spread: Real vs. fake viral

Two stories can reach the same audience. How they get there looks different, and those differences are the signals.

What we mean by viral. Spread from person to person, where each account passes a story to a few others and they pass it on. The opposite is broadcast, where many people get it straight from one source (Goel et al., 2016).

Illustrative model. Not a measurement of any particular campaign. The time axis is relative and has no units.

Astroturfing: Real vs. fake digital crowds

A crowd can be real or manufactured. Manufacturing one is called astroturfing: a paid or orchestrated campaign disguised as a grassroots effort (New York Attorney General, 2021).

Illustrative model. Not a measurement of any particular campaign. The time axis is relative and has no units.

Accounts: Real vs. fake accounts

Some accounts are run by software, or by one operator posing as many people. Researchers study how those accounts behave differently from real ones (DFRLab, 2017; Shao et al., 2018).

Illustrative model. Not a measurement of any particular campaign. Shapes, counts and timing are model devices.

Sources

• AcademicMannocci, Mazza, Monreale, Tesconi, Cresci. Detection and Characterization of Coordinated Online Behavior: A Survey (2026)Time windows as the core filter, short versus long windows, and the authentic and inauthentic taxonomy.
• AcademicPacheco et al. Uncovering Coordinated Networks on Social Media (arXiv 2001.05658)Coordination networks from shared behavior. The pump-and-dump case used 30-minute time bins and an 8-tweet minimum.
• AcademicGiglietto, Righetti, Rossi, Marino. It takes a village to manipulate the media: coordinated link sharing behavior during 2018 and 2019 Italian elections (Information, Communication & Society, 2020)Defines coordinated link sharing as repeated sharing of the same URLs within a short time, with a data-driven window.
• AcademicZannettou et al. The Web Centipede (2017)How news URLs move between Twitter, Reddit and 4chan, including uneven lags and the share that stay on one platform.
• AcademicCinus, Minici, Luceri, Ferrara. Exposing Cross-Platform Coordinated Inauthentic Activity in the Run-Up to the 2024 U.S. Election (2025)Link-based and text-based coordination across X, Facebook and Telegram, with bridge accounts between networks.
• AcademicVosoughi, Roy, Aral. The spread of true and false news online (Science, 2018)False stories spread faster and farther, driven mainly by people. Used here to show that speed alone is not a signal.
• AcademicGoel, Anderson, Hofman, Watts. The Structural Virality of Online Diffusion (Management Science, 2016)Defines viral spread as person-to-person, in many generations, versus broadcast from one source. Studied about a billion Twitter diffusion events and found structural virality typically low.
• AcademicLerman. Information is not a Virus, and Other Consequences of Human Cognitive Limits (arXiv, 2016)Argues the virus comparison has limits, because people must find a story and choose to share it.
• ReportSchliebs, Bailey, Bright, Howard. New study finds coordinated amplification network promoting tweets by China's diplomats in the UK (Oxford Internet Institute, 2021)Found 62 accounts that amplified tweets from the PRC ambassador to the UK and the Chinese Embassy in London through retweets and replies, often within sixty seconds of another account in the network. The link goes to a press release.
• ReportPew Research Center. Public Comments to the Federal Communications Commission About Net Neutrality Contain Many Inaccuracies and Duplicates (2017)Of 21.7 million comments, 6% were unique. The most-submitted comment appeared 2.8 million times, and one burst of 475,482 comments arrived at a single timestamp. Also found a real surge after a TV segment.
• ReportNew York Attorney General. Attorney General James Issues Report Detailing Millions of Fake Comments (2021)Found nearly 18 million of more than 22 million comments to the FCC were fake, and describes a paid campaign disguised as a grassroots effort. The link goes to a press release that summarizes the report.
• AcademicKeller, Schoch, Stier, Yang. Political Astroturfing on Twitter: How to Coordinate a Disinformation Campaign (Political Communication, 2020)A documented South Korean campaign. Traces of coordination among accounts separated campaign accounts from regular users better than individual account traits did. The campaign had limited influence on political discussion.
• AcademicSchäfer, Evert, Heinrich. Japan's 2014 General Election: Political Bots, Right-Wing Internet Activism, and Prime Minister Shinzō Abe's Hidden Nationalist Agenda (Big Data, 2017)Over 540,000 tweets around the election. Found bots repeatedly posting nearly identical messages, and discusses detecting near-duplicate posts.
• ReportGraham, Bruns, Zhu, Campbell. Like a Virus: The coordinated spread of coronavirus disinformation (2020, for the Centre for Responsible Technology)Early findings of a small, sustained, coordinated network of Twitter accounts pushing a bioweapon claim. Preliminary results; the link goes to a summary page.
• ToolCoordination Network ToolkitOpen-source detection of co-link, co-tweet, co-similarity and other coordination types, with a 60-second default window.
• ToolHoaxy, Indiana University Observatory on Social Media (2016, ScienceDaily)Visualizes how claims in the news, and fact-checks of those claims, spread online through social networks. The link goes to a news summary of the launch.
• AcademicTörnberg. Echo chambers and viral misinformation: Modeling fake news as complex contagion (PLOS ONE, 2018)Uses the term viral spread for the rapid spread of misinformation through a social network, and models how network structure affects it.
• AcademicBak-Coleman, Kennedy, Wack, Beers, Schafer, Spiro, Starbird, West. Combining interventions to reduce the spread of viral misinformation (Nature Human Behaviour, 2022)Uses the term viral spread for the rapid early growth of a misinformation story, and studies how interventions can disrupt it.
• ReportDFRLab. #BotSpot: Twelve Ways to Spot a Bot (2017)Treats about 72 posts a day, one every ten minutes for twelve hours at a stretch, as suspicious, and more than 144 a day as highly suspicious. Cites the Oxford Internet Institute's threshold of more than 50 a day.
• AcademicShao, Ciampaglia, Varol, Yang, Flammini, Menczer. The spread of low-credibility content by social bots (Nature Communications, 2018)Analyzed 14 million messages about 400,000 articles. Found bots especially active in the earliest moments of an article's spread, targeting users with many followers through replies and mentions.
• AcademicZouzou, Varol. Unsupervised detection of coordinated fake-follower campaigns on social media (EPJ Data Science, 2024)Detects fake-follower groups partly by finding accounts created on similar dates. Notes earlier work finding that accounts created in bursts behaved alike and were more likely to be bots.
• AcademicBellutta, Carley. Investigating coordinated account creation using burst detection and network analysis (Journal of Big Data, 2023)Studied 541 million tweets from 21 million users about the 2020 U.S. elections. Accounts created on burst days had somewhat more bots (about 1.6 to 2.2 percent more) and were more similar in behavior. The authors say the largest bursts probably include substantial real activity and that creation date alone cannot identify malicious accounts.
• AcademicLuceri, Salkar, Balasubramanian, Pinto, Sun, Ferrara. Coordinated Inauthentic Behavior on TikTok: Challenges and Opportunities for Detection in a Video-First Ecosystem (arXiv, 2025)Analyzed about 1.35 million TikTok videos from 362,000 users, August to October 2024, on the U.S. election. Found clusters of accounts posting within five-minute intervals, one cluster mostly between 6 and 8 pm UTC, and few suspicious accounts removed. This paper does not report account creation dates.

Sources checked 4 Nov 2026. Some links go to abstracts or summary pages.