The wonderful world of graphing 4chan

· Experiment

Why bother?

At university, I took an unusual path that weaved its way through sociology and computer science. Okay, that’s a fanciful way of saying I did a double degree. Applaud whenever you’re ready. Because I was intensely interested in both halves of my academia, I tried to bring them closer together instead of seeing them as separate specialisations. One way that happened was through enrolling in somewhat unpopular courses like ones centered in online research methods.

One of the assignments in that course was to analyse a web-based social medium (i.e.: social media platform) and get some insight on it. The possible choices were reddit, Twitter, or hyperlink networks (interconnected websites). Being that I’m usually that student, I opted for option D, none of the above, and went to 4chan. The problem with 4chan is that there aren’t really any tools to conduct SNA (social network analysis) on it. reddit, Twitter, and hyperlink networks all have VosonDASH, the incredible tool the course convenor, Robert Ackland, helped put together. VosonDASH does very useful things like make cool graphs out of all those social media where you can see how individual commenters or posters are linked to each other. To do that, VosonDASH creates a GraphML file, which is basically a way of describing a graph in a computer-readable way.

As they say, the best artists steal, so I took the same approach to make 4chan threads into visualisable graphs, and wrote a 4chan JSON to GraphML converter in Python. The converter uses the 4chan API to download all the information and posts in a thread, then turns it into a graph where each post is a node, and each edge or link is a reply.

In the interests of transparency, I don’t think I’m the first to do this. Lessons of Dankness was actually where I got my inspiration from, but unfortunately, they never made their code available (as far as I could see). I’ve got some example graphs made with the converter below, but for more of the same, you should absolutely check out Lessons in Dankness’s post on visualising 4chan threads. It’s based and redpilled.

Some visualisations I’ve made

All documentation should include examples, so here ya go. These were made using R and the igraph package. All of these threads were gathered as part of the aforementioned assignment, so they all discuss similar topics.

Here’s a plot of a thread discussing The Gamechangers, a vegan bodybuilding documentary:

Gamechangers visualisation

Here’s one urging ‘vegan chads’ to rise up:

Vegan chads visualisation

And here’s one discussing whether veganism is inherently flawed:

Veganism sentiment visualisation

In case you’re interested, here are the same plots with each post (node) coloured according to their ‘sentiment score’, which measures how nice or naughty the words in the post are. Orange is nicer, blue is naughtier.

Here’s the Gamechangers one:

Gamechangers sentiment visualisation

Vegan chads:

Vegan chads sentiment visualisation

And the veganism discussion:

Veganism discussion sentiment visualisation

This post brought to you by…

If you’d like to see the actual JSON and GraphML files, they’re all available by clicking or tapping these links:

Gamechangers

Vegan chads

Flawed veganism