Analysis of /pol/ and How Posting Habits Relate to Page Rank and the System of Hubs and Authorities
In a recently released academic paper (which can be found here) titled “A Longitudinal Measurement Study of 4chan’s Politically Incorrect Forum and its Effect on the Web,” researchers analyze data collected from over 8 million posts on 4chan’s /pol/ discussion board. For those that are not aware, 4chan is a popular anonymous imageboard website that hosts boards for discussion of various hobbies and interests. 4chan’s /pol/ board, officially titled “Politically Incorrect,” is described by the board’s stickied post as a board for “the discussion of news, world events, political issues, and other related topics.” /pol/ was created in October 2011 as a replacement to /new/, the site’s news board, which had been removed at the start of 2011 due to the high volume of racist discussion. In the 5 years since /pol/’s creation, the board has built up a reputation across the internet as a cesspit of racist, sexist, antisemitic, homophobic, islamophobic, etc. discussion. In recent years, /pol/ has become associated by the mainstream media with the alt-right movement and has somehow managing to force its way in mainstream news coverage on several occasions. While there is, of course, more to /pol/ than just alt-right discussion and hate speech, /pol/ can still be fairly accurately summarized as such. With this in mind, we can continue to discussion of the paper.
In the paper “A Longitudinal Measurement Study of 4chan’s Politically Incorrect Forum and its Effect on the Web,” researchers analyze data collected from over 8 million posts made on /pol/ from June 30 to September 12 of this year. Topics of interest include posting behavior of different countries, uniqueness of uploaded content, popularity of linked websites, popularity of specific hate speech by country, and effects of /pol/ users on other websites. An additional point of interest, the paper includes a small collection of impressively rare Pepes after its conclusion. While much of the data discussed in this paper is quite interesting and some of it quite humorous, I would like to specifically discuss the websites linked to by posts on /pol/ (discussed in section 4.1.1 on page 6 of the linked paper).
The study found that the top 10 most linked to websites on /pol/ were, in decreasing order, YouTube, Wikipedia, Twitter, Archive.is, WikiLeaks, Pastebin, DonaldJTrump.com, DailyMail, Beitbart, and Archive.org. Most of these websites, such as YouTube, Wikipedia, Twitter, Archive.is, WikiLeaks, Pastebin, and Archive.org are unsurprising. YouTube allows users to share videos with each other. Wikipedia is an easily accessible source of information. Twitter provides live updates and information from political events that /pol/ users would be discussing. Archive.is, WikiLeaks, Pastebin, and Archive.org allow users to dump and store information somewhat permanently in a place that is easily accessible to other users. The websites of real note here are DonaldJTrump.com, DailyMail, and Beitbart. It is, as the paper points out, quite telling that /pol/ posts link to Trump’s own website and disreputable new sites significantly more frequently than they link to news sources that are generally considered reputable or legitimate.
This relates to the page rank system of hubs and authorities that we have been discussing in class. If we describe /pol/ as a system of hubs and authorities, individual posts on /pol/ act as hubs of sorts and the websites that are linked to by posts on /pol/ act as authorities. We can consider those news sources that are linked to most frequently to be greater quality authorities in the eyes of a typical /pol/ user. This shows an interesting shortcoming of page rank and the model of hubs and authorities: these models don’t always favor the best authorities, but rather favor the most discussed or most referenced authorities. In a small-ish community (several million unique daily visitors and ~3 million posts per month in this case) or a community which shares nearly homogeneous beliefs can cause authorities which are not necessarily “good” to become ranked highly.
In case the above hyperlink doesn’t work, the paper can be found at https://arxiv.org/pdf/1610.03452v1.pdf
It is an interesting read and can be somewhat humorous for those familiar with the website.