Rudolf Nanne
Rudolf Nanne
That looks great. I find that kind of knowledge graph fascinating to look at — I used to love playing with it, for instance in Roam Research and later Obsidian. But what I missed — and I find most useful for creative work — is the ability to build and manipulate those relationships spatially, building the relationships by laying out and reordering ideas and images, as if I was working with cards, pictures and strings on a corkboard or post-its and markers on a whiteboard.
We think with hands and space as much as with words. This is why I prefer the Heptabase implementation at the moment, even if it does not have a full knowledge graph representation as such — prioritising build and manipulation over dynamic representation — although it has the options to visualise links and backlinks as lines between elements.
I think it's a key aspect of tools for thought, and given the interest this topics has among its users, it would be great to hear what the Tana team thinks about it.
I think I’m on the same page here..
A dynamic view of the graph is nice , but it’s practical use is very limited.
First of all, it only addresses consuming your existing knowledge graph.
Second, even that functionality is severely limited in the current implementations.
Getting more from an existing graph demands more then bubbles connected by tensioned wires. You need a(n)underlying Graph Query Language to select / easily filter relevant parts . You want scenarios/ stored perspectives. You want side-by-side views on parts of the graph, etc.
What we currently see is just a gimmick.
Not being able to create new information by manipulating elements of a graph by means of their graphical representations is a missed opportunity!!
Many of the best ideas / discoveries/insights comes from visual thinking and as such is often expressed first by means of graphic elements.
One area that methodologically will need attention is how to introduce more semantic value as we move from very informal drawings to more formalised ideas.
I think it would be beneficial to allow rather loosely defined semantic types at the one end , and then when concepts have been validated (on their merit), go through some stages of semantic refinement,
Richer semantics allow for more & precise reuse of knowledge, but should not be a burden to implement, nor hinder communication!
As an extreme example of such a formalised idea you might want to look at one of the ontology graphs that have been developed/evolved in the broader context of the semantic web.