Frequently asked questions about thinking, epistemology, and cognitive tools. 9738 answers
Treating the graph as the knowledge itself. The graph is a map, not the territory. You can build an elaborate, beautifully connected knowledge graph and still not understand the material it represents. The danger is spending more time maintaining the graph than engaging with the ideas. A graph.
Individual atoms of knowledge become powerful when linked into a navigable structure.
Concepts are nodes and relationships are edges — together they form a graph.
Your externalized thoughts are the raw material for a knowledge graph.
Your externalized thoughts are the raw material for a knowledge graph.
Open your primary note system. Pick 10 notes at random — not your best ones, just 10. For each note, write one sentence answering: 'What single idea does this note contain?' If you can't answer in one sentence, the note contains multiple potential nodes and needs splitting. If the sentence is.
Treating your existing notes as already graph-ready without inspection. Most notes are too long, too vague, or too tangled to function as nodes. They contain three ideas mashed together, or they summarize a source without stating your own position, or they use language so context-dependent that.
Your externalized thoughts are the raw material for a knowledge graph.
Relationships between ideas deserve as much attention as the ideas themselves.
Relationships between ideas deserve as much attention as the ideas themselves.
Relationships between ideas deserve as much attention as the ideas themselves.
A link labeled causes is more useful than a generic link labeled related.
A link labeled causes is more useful than a generic link labeled related.
Open your note system and find five links between notes. For each one, write a one-word label that describes the relationship: causes, contradicts, extends, supports, exemplifies, enables, refines, or something domain-specific. If you cannot name the relationship, ask yourself whether the link is.
Creating a link taxonomy so elaborate that you spend more time classifying relationships than building knowledge. The goal is not perfect ontological coverage. It is having enough type information that traversing a link tells you something the link's mere existence would not. Five to seven edge.
A link labeled causes is more useful than a generic link labeled related.
When A links to B, B should know that A links to it — bidirectional linking reveals hidden patterns.
When A links to B, B should know that A links to it — bidirectional linking reveals hidden patterns.
A densely connected area of your graph represents deep understanding.
A densely connected area of your graph represents deep understanding.
A densely connected area of your graph represents deep understanding.
Pick two subjects you know well and one you're just beginning to learn. For each, list 10 concepts from memory. Then draw the connections between them — every relationship you can articulate (causes, enables, contradicts, exemplifies, depends on). Count the edges. Calculate the density: edges.
Treating link count as a vanity metric. You can inflate density by creating shallow, meaningless connections — tagging everything with the same broad category, linking notes because they share a word rather than a concept. Density without semantic weight is noise. The test is whether you can.
A densely connected area of your graph represents deep understanding.