This site displays data of every substack newsletter (that is connected to the main network). On the graph, newsletters are sized proportionally to their subscriber count, and connected to each other with links based on each substack's recommendations page (which is also how the data is collected). Links on the graph are weighted based on gpt embeddings of article titles, so that more similar sounding substacks are grouped closer together. Nodes are colored based on category, determined by chatGPT reading the article titles, which is often wrong.
Don't see a substack you read in the search bar? It might not be reccomended by any other substacks, might not have a domain that ends in .substack.com, or might just be weird in some other way that my hastily written scraper didn't like. I'm working on it.
Made using d3.js.