Analyzing traffic from the The Gateway Pundit, a popular far-right website known for repeatedly sharing false information, we find that search engines and social media platforms are the main drivers of traffic and the site is more popular in counties that voted for Trump in 2020.
Chen, Zhouhan, Haohan Chen, Juliana Freire, Jonathan Nagler, and Joshua A. Tucker. "What We Learned About The Gateway Pundit from its Own Web Traffic Data." Workshop Proceedings of the 16th International AAAI Conference on Web and Social Media, (2022). https://doi.org/10.36190/2022.68
Jun 01, 2022
Area of Study
To mitigate the spread of false news, researchers need to understand who visits low-quality news sites, what brings people to those sites, and what content they prefer to consume. Due to challenges in observing most direct website traffic, existing research primarily relies on alternative data sources, such as engagement signals from social media posts. However, such signals are at best only proxies for actual website visits. During an audit of far-right news websites, we discovered that The Gateway Pundit (TGP) has made its web traffic data publicly available, giving us a rare opportunity to understand what news pages people actually visit. We collected 68 million web traffic visits to the site over a one-month period and analyzed how people consume news via multiple features. Our referral analysis shows that search engines and social media platforms are the main drivers of traffic; our geo-location analysis reveals that TGP is more popular in counties where more people voted for Trump in 2020. In terms of content, topics related to 2020 US presidential election and 2021 US capital riot have the highest average number of visits. We also use these data to quantify to what degree social media engagement signals correlate with actual web visit counts. To do so, we collect Facebook and Twitter posts with URLs from TGP during the same time period. We show that all engagement signals positively correlate with web visit counts, but with varying correlation strengths. For example, total interaction on Facebook correlates better than Twitter retweet count. Our insights can also help researchers choose the right metrics when they measure the impact of news URLs on social media.
Measuring users’ consumption of websites containing false news has remained challenging for researchers because of difficulties quantifying the spread of information as well as a lack of web traffic data access. After conducting an audit of popular far-right news websites, however, we found that The Gateway Pundit’s site traffic was publicly available. TGP not only has one of the largest percentages of web traffic among right-wing news sites but is also a highly influential publisher of large amounts of misinformation, making it an ideal case for studying fake news consumption online. Using TGP’s entire server side web traffic data for a one-month period, we were able to understand how people land on TGP, which pages they like to visit, and what is the correlation between social media engagement signal and web traffic volume.
After identifying that TGP openly collects and publishes detailed visitor traffic, we created an automated process to download data from February 3, 2021 to March 3, 2021.In total we analyzed insights from 68 million web visits to TGP. In our overview of the month’s data, we analyzed referrer links to better understand which sites send users to TGP. Then we used geo-location information to look into whether TGP visitors were more likely to be in areas that voted for Trump in the 2020 U.S. presidential election. We also clustered articles into groups in order to measure which kinds of articles were more likely to go viral. Finally, we compared this web traffic data with social media signals to examine the relationship between social media sharing behavior and false news consumption.
Our analysis found that search engines and social media sites were the main drivers of traffic to TGP, but in different ways. While search engines accounted for 88.5 percent of external referential traffic to TGP’s home page, social media sites typically linked to article pages. The top two referrers to TGP article pages were actually internal traffic, meaning that most readers clicked on articles from TGP’s home page or another article page. But after excluding internal traffic we found that social media platforms accounted for 42 percent of external referral traffic to article pages. This includes traditional social media platforms like Twitter and Facebook, as well as emerging alt-right platforms such as Telegram and Gab. Additional referrers to article pages included conservative news sites such as protrumpnews.com and search engines such as Google and Duckduckgo.
By leveraging IP addresses and city-level geolocation labels in the web traffic records, we also found that visitors to TGP were in fact more likely to be from areas that voted for Donald Trump in the 2020 election. While this finding supports our initial hypothesis, it is important to note that this is observational data and does not mean we can infer a causal relationship between site visits and support for Trump. From here we analyzed topic groups for articles and found that pieces related to “election fraud” and “capital riot” received more visits than other topics.
Taken together, our analysis shows that social media platforms and search engines play a significant role in referring people to information, emphasizing the importance of transparency from platforms when it comes to outgoing web traffic, not just internal metrics such as retweets. In future studies, we hope to collaborate with industry partners that have direct access to outgoing web traffic data in order to further our understanding of this relationship.