4 comments

[ 4.6 ms ] story [ 15.8 ms ] thread
Each dot is a subreddit. "countries" are clusters of usually related subreddits. Relationship is determined based on Jaccard Similarity (users who post to A also post to B)

This was my hobby over past few years. I collected ~176 million unique pairs `(user, subreddit)` of comments, computed jaccard similarity, clustered a graph, performed a layout, created a webgl renderer and streaming svg parser, and today releasing it to the wild. Hope you enjoy it!

* The source code for the rendering is available here: https://github.com/anvaka/map-of-reddit

* To change the map - go here https://github.com/anvaka/map-of-reddit-data

I'm assuming the white dots are also connected subreddits but not enough to show a line?

Most of the results I've tried are exactly what I expected e.g. users from my alma mater also post to the same regional subreddits. Were you able to find any interesting/unexpected connections?

I'm working on a similar product
This is really well done, nice job! It's informative and fun to explore.