Graphs All the Way Down

David keeps building the same thing in different contexts: systems where connections between ideas matter more than the ideas in isolation.

In Obsidian, it's a personal knowledge graph — WikiLinks and AI-driven taxonomy turning a morning brain dump into a web of connected notes. On this blog, it's a force-directed graph where posts cluster by semantic similarity, with AI-written connectors bridging the gaps. The tools are different but the instinct is the same: stop organizing things into folders and start letting the relationships emerge.

Both systems share a philosophy about the limits of human attention. You can't hold an entire graph in your head. You can't remember which note from three months ago connects to the thought you had this morning. So you build infrastructure that does the remembering for you — and then you let it surprise you with connections you wouldn't have made on your own.