thimkin bout graph layout again
@jonny how big a graph? :o
@mauve
that seems to be the problem huh
@mauve
I am like how come node identity is always coded as being in a particular location. if you had a notation for indicating "portals" in the graph where a bunch of nodes that would appear in two otherwise distant clusters can be in two places at the same time?
so that's a multi-scale question, because then you would want the second order clusters nearby, so you'd really need to think about laying out clusters of clusters, and the distinction of scale shouldn't have to be something you pick once but scale to the granularity you're evaluating.
then thinking about the spring algos for layout, a portaled relationship could be modeled in the elastic system too, and I'm thinking about how it could change when things are in multiple places at once
@mauve
I am of course as always thinking about graph traversal, so you could think about this like a DHT. where as a node I have a linear position in hash space, but eg. kademlia constructs a graph to the line where I am connected to other nodes in k bins.
but like what if a node had multiple locations, or, another way, a peer could be represented by multiple positions in hash space. that might be the case with shared identity, where many peers make an agreement to behave as one in certain ways. so "what up you can find me here and here and here"
and so the layout question is in part thinking about how to layout such graph communities for exploration. so one might want to emphasize one particular local graph region as the thing you're evaluating for its multiple clusters across different versions of the same identity.
basically I want to be able to surf the swarm idk I want the good interface yno
visualizing both the origin of data and the interconnectedness of data seems like a hard to challenge.
have you tried drawing up some interface ideas already?
maybe having two modes where you start by exploring the data as is but then can toggle seeing where the data comes from and a separate network graph for interconnected peers be useful
