Part 01 — Insight Artifact
The Non-Obviousness
Standard
The central challenge of Part 01 is finding a concept pair that is genuinely non-obvious — not merely underdiscussed, but requiring the AI to reason across domains rather than recall a documented connection. The checker evaluates three dimensions automatically. This page explains what each one means and why it matters.
Low Co-occurrence
The two concepts must rarely appear in the same context. The question isn't whether the connection exists somewhere — it's whether it's sparse enough that an AI can't simply retrieve and repeat it.
If there's a Wikipedia article, a well-known essay, a TED talk, or a canonical piece of writing that already makes this connection explicitly, the pair fails this criterion. The connection must exist in the gap between documented knowledge, not within it.
Cross-domain Distance
The two concepts must come from genuinely different fields or domains. Connecting them must require the AI to reason by analogy — finding structural parallels between worlds that don't naturally communicate — not recall an established association.
"Marketing and advertising" are adjacent, not cross-domain. "Byzantine military logistics and containerized software deployment" are. The further apart the fields, the more the AI has to actually think rather than remember.
Non-substitutable with Search
A smart person could not reproduce the core insight by searching and copy-pasting existing sources. The artifact must require synthesis — iterative reasoning that couldn't exist as a search result.
If the key idea is already sitting in an article somewhere, the artifact is retrieval, not creation. The test is whether the work could only have been produced through a genuine back-and-forth with the AI — building something that didn't exist before.
How scoring works
Each dimension is scored independently. Co-occurrence and cross-domain distance are each worth up to 4 points; substitutability is worth up to 2 points, for a total of 10. A pair needs a score of 7 or higher with no dimension at zero to pass. A score of 5–6 is borderline; below 5, or any dimension scoring zero, is a fail.
Examples
| Concept Pair | Why it fails or passes | |
|---|---|---|
| ✕ | Mental health + social media | Deeply co-documented; thousands of academic papers and popular articles already make this connection explicitly |
| ✕ | Climate change + extreme weather | Same domain; the causal link is established scientific consensus with enormous co-occurrence |
| ✕ | Leadership + sports coaching | Heavily documented; the genre of "sports metaphors for business leadership" is a well-worn category |
| ✓ | Roman aqueduct engineering + SaaS churn mechanics | Sparse co-occurrence; connecting them requires structural analogy across 2,000 years and two entirely unrelated fields |
| ✓ | Competitive rowing technique + distributed team coordination | Low training association; the insight requires genuine lateral reasoning about synchronization and feedback loops that doesn't exist pre-formed |
Run your concepts through the checker to get a score and, if you pass, your approval code.