This is neat, however (here comes the harsh critic) :
- Rendering a simple tree like this would work just as well, if better, in text only format (like the output of the "tree" command on linux) : https://justpaste.it/6r9un
- The classification also seems a bit manual and arbitrary.
Maybe giving each node a weight would make d3 worth using ?
Maybe scrap the subscribers of each subreddit, and make a force directed "map of reddit nsfw" ?
Your skills seem wasted on this if you're good at d3
Your text-based tree is nice, but it comes with pros and cons, as every datq visualization. In fact, it was my first idea, but I felt I need to create something better for exploring the high-level structure.
If you wanted to do more than just a straight tree: For any pair of subreddits draw a line between the subs weighted according to the number of commenters that commented in both subreddits. Maybe doing a similar graph with people that post in both subreddits would be interesting too. I expect both variants to yield a somewhat more natural grouping of subreddits than an artifically curated list, but making it readable and explorable would be a challenge.
If I were to do so, I know more suitable methods, vide description from https://github.com/stared/tagoverflow/. Without such normalization, it won't work given that there are networking with different orders of magnitude of users.
Right now don't have much time for this project, and well - I am this kind of person that has a few dozen times for idea than time, not the other way. :)
- Rendering a simple tree like this would work just as well, if better, in text only format (like the output of the "tree" command on linux) : https://justpaste.it/6r9un
- The classification also seems a bit manual and arbitrary.
Maybe giving each node a weight would make d3 worth using ? Maybe scrap the subscribers of each subreddit, and make a force directed "map of reddit nsfw" ?
Your skills seem wasted on this if you're good at d3