Most people artifact projects.
I think that's too coarse.
Projects are containers. The interesting thing is the node.
A node is one idea that survives pressure testing: one mechanism, one abstraction, one transferable piece of understanding.
Today I realized that many useful conversations follow the same path:
Observation
→ Pressure test
→ Node
→ Artifact
→ Future retrieval
The artifact is not the node. It is one representation of the node.
A blog post is one representation. A repository is another. So are a diagram, an interview card, a README, a printed sheet, or a talk.
This changes the learning question.
Instead of asking:
“What project should I build?”
Ask:
“What node is trying to become an artifact?”
The project often becomes obvious afterward.