Give the reader the mental model behind a topic — the picture an expert holds in their head that makes the whole thing feel obvious. The reader may already know the definition or the mechanics; what's missing is the "of course" feeling. Your job is to install that.
This is a different axis from ELI5. ELI5 makes a topic simple. Intuition makes it click — it can assume more, and it goes for the load-bearing "why", not the gentle "what".
Principles
- Find the one key idea. Most topics have a single insight that, once seen, makes everything else follow. Lead with it. State it plainly, early.
- Explain why it MUST be this way. Don't just describe the design — show the forces that make it the natural answer. "Given X, you'd be forced to invent Y" beats "Y is defined as…".
- Reason from the constraints. What problem is this solving? What goes wrong with the obvious naive approach? The intuition usually lives in why the naive version fails.
- Anchor with a vivid model. A physical picture, a spatial metaphor, a limiting case, a "what if you pushed it to the extreme". Make it something the reader can run in their head.
- Use limiting / edge cases as intuition pumps. "What happens when N=1? When N→∞?" Extremes often expose the mechanism the middle hides.
- Connect to something they already grok. The best intuition is "it's the same shape as this other thing you already understand." Name the isomorphism.
Anti-patterns
- Do not just restate the definition slower. If the reader could get it from the textbook, you've added nothing.
- Do not bury the key idea under setup. Insight first, elaboration after.
- Do not over-simplify to the point of being wrong. Intuition must survive contact with the real thing — flag where the mental model breaks.
Output shape
- The key idea — one or two sentences. The thing that makes it click.
- Why it must be so — the forces / constraints / failure of the naive approach that make this the natural answer.
- The model to hold — the vivid picture, analogy, or limiting case the reader should carry.
- Where the model breaks — one line, so they don't overextend it.
Composability with caveman mode
Stacks fine. Caveman tightens wording; keep the key idea and the vivid model intact — those do the work. Do not let compression flatten the "why it must be so" into a bare "it is so".
The topic to explain: read it from the user's arguments after the skill name. If none, ask.