I scanned the headlines but nothing jumped out.
When you are zoomed out on the graph view, all you can see are the vague constellations. To see the labels on the vertices requires monotonous zooming in and zooming out.
Add a magnification “call-out” (like in an architectural blueprint) that shows the nodes under the mouse pointer so you can click on nodes without having to zoom in or guess what they are.
I was thinking about the issue of a huge graph aswell recently. Now your solution seems like it would add some value to the graph feature for sure.
As I read your post another potential feature that could alleviate this occurred to me. Imagine that the graphs were organized into layers by Obsidian. There would be some algorithm that according to the current zoom level makes a selection of nodes based on how many other nodes are connected to them. Nodes with a similar amount of nodes connected to them would be shown simulataneously up to a certain extent. Now because of the complexity of huge graphs this would probably also require putting nodes into some kind of queue that would yield nodes when other nodes of the same “organization type” have been faded in and out. This way you could have some kind of “meta-zoom” where only one layer with a navigatable amount of nodes would be shown but you could still zoom through every single node without cluttering the view.
Now I could also see your proposed feature blend into that nicely. Imagine that you “meta-zoomed” out and see one of the upper layers but you think “Hmm, this is nice but I’d like to see more of where that particular nodes blends into (which you now see the label of because of the ‘meta-zoom’).”. Now you would use a tool to restrict the “meta-zoom” feature to that particular node or area of nodes and as you zoom in and out of that selected part of the view deeper layers of that particular area reveal themselves. As soon as you were satisfied you would release the restriction and the view would snap back to the standard layer.
That would make the graph view the ultimate navigation tool in my opinion. I also realize that this probably requires lots of development if possible at all.
What do you think?
I would think that adding the layers to the graph view as another dimension would certainly be a lot of work–not that it wouldn’t be cool. Additionally, an area coloring that surrounded nodes that were highly linked from a central node would be less “messy” than a node spiked with lots of edges.
I figure that they already have this graph drawing feature, it wouldn’t be super hard to add a magnified detail view. Then later they could explore the layers or areas vizualizations.
Sounds promising! I mean at this point the graph view would be like a heavily slowed down version of the actual thinking process in the brain. Amazing.
Workaround: The graph display has a threshold that hides text labels. Turn it to a low setting and poof! No more mystery nodes.