Source-linked AI summary
The Web Centipede: Understanding How Web Communities Influence Each Other Through the Lens of Mainstream and Alternative News Sources
Savvas Zannettou, Tristan Caulfield, Emiliano De Cristofaro, Nicolas Kourtellis, Ilias Leontiadis, Michael Sirivianos, Gianluca Stringhini, Jeremy Blackburn
TL;DR
The paper addresses the limited holistic evidence on how mainstream and alternative news spread across interconnected Web communities. It measures URL-sharing and cross-platform influence across Twitter, Reddit, and 4chan, finding that Twitter is highly influential overall while The_Donald and /pol/ are especially influential for alternative news. The analysis is limited because it tracks URLs rather than the content of stories and can be affected by missing Twitter data.
Problem
Little prior work provides a holistic view of how mainstream and alternative news proliferate across the modern information ecosystem, limiting understanding of alternative news dissemination.
Method
The paper analyzes millions of posts from Twitter, Reddit, and 4chan and models cross-platform influence using Hawkes processes.
Results
Twitter has the greatest influence on the number of URLs posted elsewhere, while The_Donald and /pol/ are the next most influential sources for alternative news URLs.
Takeaways & Limitations
The findings provide evidence that fringe communities can spread alternative news to mainstream social networks and inform systems for detecting, tracing, and mitigating false information.
Takeaways & Limitations
The analysis tracks URLs rather than the content of shared information and may be affected by missing Twitter URLs.
Abstract
from arXiv · showhide
As the number and the diversity of news outlets on the Web grow, so does the opportunity for "alternative" sources of information to emerge. Using large social networks like Twitter and Facebook, misleading, false, or agenda-driven information can quickly and seamlessly spread online, deceiving people or influencing their opinions. Also, the increased engagement of tightly knit communities, such as Reddit and 4chan, further compounds the problem, as their users initiate and propagate alternative information, not only within their own communities, but also to different ones as well as various social media. In fact, these platforms have become an important piece of the modern information ecosystem, which, thus far, has not been studied as a whole. In this paper, we begin to fill this gap by studying mainstream and alternative news shared on Twitter, Reddit, and 4chan. By analyzing millions of posts around several axes, we measure how mainstream and alternative news flows between these platforms. Our results indicate that alt-right communities within 4chan and Reddit can have a surprising level of influence on Twitter, providing evidence that "fringe" communities often succeed in spreading alternative news to mainstream social networks and the greater Web.
1 INTRODUCTION
Alternative narratives spread widely online, but how they move across the modern information ecosystem remains poorly understood. This paper studies mainstream and alternative news flows across Reddit, Twitter, and 4chan using large-scale measurements and influence modeling.
- Motivation: The Web has enabled conspiracy theories, false stories, harassment campaigns, and other alternative narratives to spread broadly.Examples include false claims about the Boston Marathon bombings, GamerGate, and Pizzagate.
- Research gap: Although prior work examined cascades, rumors, and hoaxes, little research provides a holistic view of how alternative narratives proliferate across the Web.The authors identify this gap as important for understanding alternative news and designing detection or mitigation strategies.
- Approach: The paper measures mainstream and alternative news flows across Reddit, Twitter, and 4chan, selected for their distinct structures and substantial online influence.The study focuses on how these platforms and their sub-communities generate, incubate, and disseminate information.
- Contributions: The study compares news-source occurrence, temporal posting dynamics, and influence between platforms to characterize information spreading across the Web.Its contributions combine large-scale measurement with analysis of when URLs appear and how activity on one platform relates to activity on another.
- Overview of findings: Twitter, Reddit, and 4chan are used extensively to disseminate both mainstream and alternative news, with platform influence varying by news type.The paper uses Hawkes processes to model these cross-platform influence relationships.
2 BACKGROUND AND DATA COLLECTION
The study examines Twitter, Reddit, and 4chan using platform-specific datasets and a curated set of 99 mainstream and alternative news sites. The datasets capture large-scale URL-sharing activity, with alternative news appearing especially often on 4chan and within selected Reddit communities.
- Platforms: Twitter is a directed micro-blogging network where users broadcast short tweets, use hashtags, and rebroadcast messages through retweets.These features support topic discovery and message diffusion among followers.
- Platforms: Reddit is a social news aggregator where users submit URLs, vote on posts and comments, and discuss them in threaded communities.Voting determines content ranking, while community structure is not defined primarily by friendship relations.
- Platforms: 4chan is an anonymous imageboard organized into topic-specific boards, with threads that are deleted after seven days and limited social-network features.The study focuses on /pol/ and uses /sp/, /int/, and /sci/ as comparison boards.
- News sources: The news-source corpus contains 99 sites: 45 classified as mainstream and 54 as alternative, including Sputnik News and RT as state-sponsored alternative domains.Posts, threads, and comments containing URLs from these sites were collected across the three platforms.
- Dataset characterization: Alternative news URLs were about twice as likely to appear in 4chan posts as in Twitter or Reddit posts.Mainstream news URLs occurred in a greater percentage of posts on 4chan and Reddit than on Twitter.
- Datasets: The collection includes 487k tweets, approximately 1.8M Reddit posts/comments containing source URLs, and 97k 4chan posts and replies.The platforms were observed from June 30, 2016 through February 28, 2017, with noted collection gaps for 4chan.
3 GENERAL CHARACTERIZATION
This section characterizes mainstream and alternative news across Twitter, Reddit, and 4chan by examining domains, URL repetition, and user-level sharing patterns. Alternative news is especially prominent among Twitter users and in selected Reddit and 4chan communities, with platform-specific differences in outlet popularity.
- Reddit: 76k alternative-news URLs and 600k mainstream-news URLs were found across six selected subreddits, with the top 20 domains covering 99% and 89% of their respective categories.Known alt-right and state-sponsored alternative outlets were predominant among the alternative domains.
- Cross-platform domain prevalence: 129k alternative-news URLs and 42k unique URLs were found on Twitter, while /pol/ contained 21k alternative-news and 82k mainstream-news URLs.The top domains covered 86% of Twitter URLs and 87% and 99% of mainstream and alternative /pol/ URLs, respectively.
- URL repetition: On Twitter, alternative URLs tend to appear more often than mainstream URLs, whereas /pol/ and the six subreddits show similar repetition for both categories.Across platforms, a substantial portion of URLs appears only once.
- Cross-platform domain prevalence: The four leading alternative domains—breitbart.com, rt.com, infowars.com, and sputniknews.com—have similar prominence across platforms, but other outlets are platform-specific.therealstrategy.com is popular only on Twitter, while lifezette.com and veteranstoday.com are popular on the six subreddits and /pol/.
- Interpretive caveat: Twitter bots may contribute to the platform’s distinctive alternative-news patterns because bots are more prevalent on Twitter than on Reddit and possibly 4chan.The authors note that bot activity cannot be excluded with certainty on 4chan.
- User-level sharing: 80% of Twitter and Reddit users share only mainstream URLs, while 13% of Twitter users exclusively share alternative URLs and mixed-category users show a wide range of alternative-news fractions.Only 5% of Twitter users who share both categories have an alternative-news fraction below 0.2, indicating comparatively frequent alternative-news sharing among these users.
4 TEMPORAL DYNAMICS
The temporal analysis compares how alternative and mainstream news URLs are reposted within platforms and first appear across platforms. It finds faster Twitter reposting, earlier appearances on selected subreddits, and platform-specific first appearances by news domain.
- Within-platform dynamics: Specific sub-communities, especially /pol/ and six selected subreddits, show much higher shares of alternative news than other communities.
- Within-platform dynamics: Twitter reposts URLs with shorter delays than Reddit and 4chan, while mainstream news propagates faster than alternative news on the six selected subreddits.The 24-hour inflection point appears for both news types across all three platforms; the mainstream–alternative difference is not evident on Twitter and /pol/.
- Within-platform dynamics: Mean inter-arrival-time distributions differ significantly between every platform pair, although /pol/ and the selected subreddits show similar sharing behavior for both news types.Two-sample Kolmogorov–Smirnov tests report p < 0.01 for each pairwise comparison.
- Cross-platform dynamics: Alternative URLs appear across platforms faster than mainstream URLs, with statistically different pairwise timing distributions and a recurring turning point near 24 hours.The pairwise distribution comparisons reject the null hypothesis with p < 10^-4.
- Cross-platform dynamics: For URLs appearing on all three platforms, R→T→4, R→4→T, and T→R→4 are the three most common sequences.Here, R denotes the six selected subreddits, T Twitter, and 4 /pol/.
- Platform influence patterns: Selected subreddits usually precede Twitter and /pol/ for both news types, while first-appearance patterns vary by domain and /pol/ does not dominate any domain.Breitbart URLs most often first appear on the selected subreddits, whereas Infowars, RT, and Sputnik URLs more often first appear on Twitter; mainstream domains also vary.
5 INFLUENCE ESTIMATION
The paper uses discrete-time Hawkes processes to estimate cross-platform influence while accounting for externally caused events. Results show that Twitter, The_Donald, and /pol/ play important but news-type-dependent roles in propagating URLs.
- Influence estimation: Hawkes processes estimate how events on one platform influence later events on other platforms while accounting for external information sources.The model includes background event rates and impulse responses between processes.
- Data preparation: A one-minute event bin preserves cross-platform timing because 92% of events occur alone and another 5.4% share bins only with events from the same platform or subreddit.The bin size is presented as a compromise between accuracy and computational cost.
- Results: Twitter has the highest self-influence weights, measuring 0.1096 for mainstream URLs and 0.1554 for alternative URLs.The paper relates this pattern to the common practice of retweeting and offers bot activity or user behavior as possible explanations for the alternative-news difference.
- Results: Twitter contributes heavily to both news types and has greater percentage influence on alternative than mainstream events across all other platforms and subreddits.The percentage estimates are derived from influence weights, source-platform event counts, and destination-platform event counts.
- Results: The_Donald and /pol/ are major alternative-news inputs to Twitter, causing estimated shares of 2.72% and 1.96%, respectively.The_Donald causes 8% of /pol/’s alternative-news URLs, while /pol/ causes 5.7% of The_Donald’s; for mainstream URLs, the direction reverses to 8.61% versus 6.13%.
- Overall implications: The_Donald and /pol/ account for around 6% of mainstream-news URLs and over 4.5% of alternative-news URLs posted to Twitter that appear across all three platforms.The paper presents this as evidence that smaller fringe communities influence Twitter and may affect the wider Web disproportionately relative to their user base.
6 RELATED WORK
Prior studies examine rumor propagation, false-information detection, source identification, and influence limitation, often within individual networks or specific cases. This paper distinguishes itself by analyzing information diffusion across multiple services and measuring influence between them.
- Disinformation dynamics: Prior work studies rumor characteristics, debunked stories, and false-information diffusion primarily through Twitter datasets, platforms, or case studies.Examples include analyses of 1.7B tweets, 1.4M tweets, and individual hoaxes or crises.
- Detecting false information sources: Other studies formulate false-information source detection using rumor centrality, reverse diffusion, monitoring nodes, or graph analysis.These approaches aim to identify likely sources or understand how websites promote conspiracy theories and political agendas.
- Contribution relative to prior work: Unlike prior work, this paper takes a comprehensive multi-service perspective on Reddit, 4chan, Twitter, and news sites.It analyzes information dynamics and flow across these services and measures influence from each platform to the others.
7 DISCUSSION & CONCLUSION
The study finds that news-sharing patterns and influence differ across Twitter, Reddit, and 4chan, with Twitter exerting broad influence and 4chan sometimes preceding later-popular stories. Its URL-based analysis offers an initial cross-platform view, but excludes information conveyed through content such as images, videos, and textual claims.
- Cross-platform findings: Users across platforms prefer different news sources, especially alternative ones, and Twitter and Reddit often post the same stories shortly before 4chan.When stories become popular after one or two days, they were usually posted on 4chan first, lending some support to its presumed Web influence.
- Cross-platform findings: Twitter has the greatest influence in causing URLs to be posted on other platforms, while platform interplay varies across mainstream and alternative news.The influence model also accounts for external information sources, revealing subtle but meaningful cross-platform effects.
- Contribution: The analysis is presented as the first attempt to characterize mainstream and alternative news dissemination across multiple social platforms and estimate influence between them.The authors position these measurements as useful for understanding and detecting false information and for designing systems that trace and mitigate fake-story dissemination.
- Limitations: The study measures propagation through URLs from mainstream and alternative news sources rather than through the news content itself.It therefore omits textual claims, images, videos, screenshots, and continued discussion without a relevant URL, limiting coverage of how information propagates.