That being said - it seems like a very cool library!
Haha, completely unnecessary. Just wanted to show off.
I've open sourced the force directed graph viz part of it in case anyone is interested: http://github.com/dhotson/springy
I would love to see competition for it as a collaborative webapp or as a native OS X / iPhone / iPad app.
Seriously. The graph is one of the best data structures we have for modeling systems and taxonomies.
See, for example, Avatar. What enabled the effects in that film, at its very core?
A Directed Acyclic Graph.
I think the library looks really interesting.
I've used it in the past to visualize the structure of Google Wave conversations, and intend to use it in the future to visualize related products. (Unless I use this one instead, of course :)
I think that a graph interface like this shouldn't be the only way to get at some set of data, but at least for some people, it can be a really new and compelling way to explore it.
I ask because it uses a Force Directed layout, I guess it will get very slow after a few thousands of nodes.
Looks promising though. I'll play with it...
Christian's site, with some of his other work, is at http://samizdat.cc/
seems as if the algorithm grows unstable for very high node repulsion. even with a few nodes you can tell it is getting caught in a local energy minimum, plus the oscillations are jarring. maybe just cap the repulsion strength at 50k?
have you thought about adding hover text, upon a mouseover of the vertices?
also, maybe instead of spring tension, one could use fixed lengths given by the weights of another text file. giving user generated meaning to the edge lengths.
(I'm working on it)