Post

im thinking... have you tried to express edge differently ? e.g. what if it was semantic similarity between people, and not just do they follow each other ?

maybe some clustering based on the interest of the account. e.g. account of a webgl dev, of a AI researcher, etc...

Maybe we could search : "who is in my follower is AI researcher" or "Who like to cook sushi".

works fine for 100-200 people, but after that it gets a bit chaotic ^^

Nice! Maybe you could only show the 25-50 folks that interacted the most with you in the past months to keep the graph manageable? And maybe for each of them show ~10 or so of their top one?

Yeah I tried something like that, but even though it was less people, it kinda lost the cool "fragmented" look where you can see different groups that kinda stick together and all follow each other