3 Million Russian Troll Tweets (FiveThirtyEight)
The 2,946,207 tweets sent from 2,843 Twitter handles that Twitter listed to the US Congress as connected to the Internet Research Agency, a Russian "troll factory", as FiveThirtyEight published them in 2018 after receiving them from the Clemson University researchers Darren Linvill and Patrick Warren, who gathered them with a social-media listening tool. Each tweet has its text, the handle, the posting time, the collection time, the account's follower and following counts at that time, a region and a language assigned by the tool, whether it is a retweet or quote tweet, the researchers' account type and category, the link to the original tweet, and the first redirects of up to three t.co links in the text. 617,434 tweets are from the 931 handles that were first listed in June 2018.
Cite this dataset at
russian-troll-tweets.echo.directory
Analytics
The full graph has 592,043 nodes; this view shows the 1,000-node sample the pipeline drew for display. Node size is centrality, colour is community.
Topic 0
Topic 1
Topic 2
Topic 3
Topic 4
Topic 5
Topic 6
Topic 7
Topic 8
Topic 9
- NonEnglish820.8K (27.9%)
- RightTroll711.7K (24.2%)
- NewsFeed598.2K (20.3%)
- LeftTroll427.1K (14.5%)
- HashtagGamer241.8K (8.2%)
- Commercial121.9K (4.1%)
- Unknown13.5K (0.5%)
- Fearmonger11.1K (0.4%)
Computed over the full corpus.
Record
- Catalog key
russian-troll-tweets- Authors
- Linvill, Darren L.; Warren, Patrick L.
- Year
- 2020
- Source paper
- Troll Factories: Manufacturing Specialized Disinformation on Twitter
- DOI / arXiv
- 10.1080/10584609.2020.1718257
- License
- CC BY 4.0 (FiveThirtyEight data) (CC-BY-4.0)
- Records
- 2,946,206
- Nodes
- 592,043
- Edges
- 832,751 (directed)
- Nodes are
- account
- Edges are
- mention (author -> @mentioned account)
- Attributes
- nodes have no attributes; edges have attributes
- Text corpus
- tweet text
- Languages
- de, en, ru
- Collected
- 2012-02-02 to 2018-05-30 (2010s)
- Tasks
- Text classification (multi-class;
troll_category_label, post_type_label) inferred
From the graph: link prediction, community detection - Region
- Global (inferred: the platform is used worldwide)
Provenance
- Originally published at
- https://github.com/fivethirtyeight/russian-troll-tweets link checked
- Original files as received
- The source files, unconverted, kept for reference: mirrored on echo.directory on 2026-10-20, 1.1 GB across 14 files. Checksums in MANIFEST.json. For use, start with the data package below.
- Data package
- Version 1.0.0 of this dataset as an ECHO Data Package: posts (2,946,207 rows).
- Structure
- Temporal (timestamped), Text
Citations
Troll Factories: Manufacturing Specialized Disinformation on Twitter
D. Linvill, Patrick L. Warren
Political Communication · 2020
Cited by
- Cross-Platform State Propaganda: Russian Trolls on Twitter and YouTube during the 2016 U.S. Presidential Election
Yevgeniy Golovchenko, C. Buntain et al. · The International Journal of Press/Politics · 2020 · cited 165× - A Confirmation Bias View on Social Media Induced Polarisation During Covid-19
Sachin Modgil, R. Singh et al. · Information Systems Frontiers · 2021 · cited 141× - Content-based features predict social media influence operations
Meysam Alizadeh, Jacob N. Shapiro et al. · Science Advances · 2020 · cited 107× - Exposure to the Russian Internet Research Agency foreign influence campaign on Twitter in the 2016 US election and its relationship to attitudes and voting behavior
G. Eady, Tom Paskhalis et al. · Nature Communications · 2023 · cited 106× - Studying fake news spreading, polarisation dynamics, and manipulation by bots: A tale of networks and language
G. Ruffo, Alfonso Semeraro et al. · Computer Science Review · 2021 · cited 102×
Show all 10
- Computers in Human Behavior
D. Linvill, Brandon C. Boatwright et al. · cited 100× - Troll and divide: the language of online polarization
Almog Simchon, W. Brady et al. · PNAS Nexus · 2020 · cited 94× - #MaskOn! #MaskOff! Digital polarization of mask-wearing in the United States during COVID-19
Jun Lang, Wesley W. Erickson et al. · PLoS ONE · 2021 · cited 92× - Fake news believability: The effects of political beliefs and espoused cultural values
Manjul Gupta, Denis Dennehy et al. · Information Manager (The) · 2022 · cited 84× - The Routledge Companion to Media Disinformation and Populism
S. Waisbord, R. Armitage et al. · 2021 · cited 82×
Citation data from Semantic Scholar, retrieved 2026-10-21.
How to cite
BibTeX
@misc{echo2026,
title = {ECHO: Exploratory Compendium of Humanity Online},
author = {Switzer, Braden and Zafarani, Reza},
year = {2026},
url = {https://echo.directory},
note = {Syracuse University}
}
@article{linvill2020troll,
title = {Troll Factories: Manufacturing Specialized Disinformation on Twitter},
author = {Linvill, Darren L. and Warren, Patrick L.},
journal = {Political Communication},
volume = {37},
number = {4},
pages = {447--467},
year = {2020},
doi = {10.1080/10584609.2020.1718257}
} APA
Switzer, B., & Zafarani, R. (2026). ECHO: Exploratory Compendium of Humanity Online [Data repository]. Syracuse University. https://echo.directory
Linvill, D. L., & Warren, P. L. (2020). 3 Million Russian Troll Tweets (FiveThirtyEight) [Data set]. https://doi.org/10.1080/10584609.2020.1718257
MLA
Switzer, Braden, and Reza Zafarani. ECHO: Exploratory Compendium of Humanity Online. Syracuse University, 2026, echo.directory.
Linvill, Darren L., and Patrick L. Warren. 3 Million Russian Troll Tweets (FiveThirtyEight). 2020, doi.org/10.1080/10584609.2020.1718257.