# Lateral Thinking > Generate non-obvious hypotheses by decomposing a problem to its mechanism and raiding distant fields for transferable patterns, beyond standard advice. Use distill first when the system itself is still unclear. - Skill: `abpai/lateral-thinking` (Agent Skill, multi-file: 3 files) - Install (CLI): `npx skillmds@latest add abpai/lateral-thinking` - Raw SKILL.md: https://api.skillmd.com/api/skills/abpai/lateral-thinking/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - License: MIT - Author: abpai (https://skillmd.com/u/abpai) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/abpai/lateral-thinking --- # Lateral Thinking Generate novel, testable ideas by finding mechanisms that transfer across fields. Use this skill when ordinary analysis is already exhausted and the user needs a good second or third lens, not a recap of the obvious first one. `distill` compresses a system to its essential primitives; `lateral-thinking` uses those primitives to generate non-obvious hypotheses. If both apply, distill first, then run lateral thinking on the clarified skeleton. ## Ring Model Start beyond the obvious. - **Ring 0-1:** standard advice, baseline literature, first-order domain answers - **Ring 2:** component decomposition and overlooked regulators - **Ring 3:** cross-domain transfer of mechanisms from distant fields Do not waste most of the response on Ring 0-1. A brief baseline is fine only if it helps make the lateral leap understandable. ## Workflow ### 1. State the problem skeleton Strip away jargon and restate the raw mechanics of the problem in 2-3 sentences. Example: "We ship features, users sample them once, then their behavior snaps back." If the framing is ambiguous, high-stakes, or likely to drift, confirm the skeleton with the user before going deeper. Otherwise, proceed with the explicit stated skeleton and note that it is your working model. ### 2. Decompose into primitives Inspect the mechanism through a few consistent lenses: - information flow - timing and sequencing - incentives - structural constraints - feedback loops - resource flows Ask: - What regulates this component that no one is watching? - What adjacent system touches it? - What happens if the sign flips? - What is the dual or inverse? This step is complete when the skeleton names the main actors, constraints, feedback loops, and a likely regulator or missing variable. ### 3. Generate Ring 2 discoveries Produce only the non-obvious observations that change the diagnosis or available actions. Do not fill a quota. ### 4. Run a cross-domain raid Search for the same mechanism in genuinely distant fields. Choose source fields because the mechanism transfers, not because they appear in a standard list. For each candidate analogy, name the mechanism that transfers. Avoid surface-level metaphors. This step is complete when every candidate names the source field, transferable mechanism, and reason it is not a decorative analogy. ### 5. Synthesize hypotheses For each promising mechanism transfer, write: - the non-obvious connection - the mechanism chain - why this idea is not already standard in the target field - what nearby evidence or adjacent literature would support it - what concrete test would falsify or validate it - the likely impact if true Keep only hypotheses distinct enough to test separately. ### 6. Check intersections Look for combinations where two hypotheses reinforce or unlock each other. Sometimes the real insight is not one borrowed mechanism, but the interaction of two borrowed mechanisms. This step is complete when each surviving hypothesis has a falsifier or concrete test, and intersections are named separately from standalone ideas. ### 7. Rank and kill Run an adversarial pass on every surviving hypothesis: - does the mechanism transfer, or only the metaphor? - do the quantitative assumptions port? - is this already known and discredited elsewhere? - does it produce an actionable next step? Kill or downrank ideas that fail these tests. Rank survivors by mechanistic plausibility, domain distance, and testability. Penalize obvious domain advice, ideas already tried, hand-wavy suggestions with no test, and ideas far outside the user's practical reach. Judge reach against the authority the user's own framing implies; ask only if the ranking turns on it. This step is complete when every hypothesis carries a verdict (survives, downranked, or killed), and the survivors are ranked. ### 8. Recommend actions Turn the top-ranked survivors into a small set of concrete next steps. Each step names what to do and which hypothesis it tests. ## Output Format Follow the template in [references/output-format.md](references/output-format.md). ## Output Choice Return Markdown by default. For workshops, strategy debates, or broad ideation sets, create a self-contained HTML hypothesis board instead of a long Markdown report. Use cards for Ring 2 discoveries and Ring 3 hypotheses, show source fields as labels, make tests/falsifiers visible, and group intersections so the user can compare candidates side by side. ## Iteration If the best hypothesis reframes the problem, run one more cycle with the updated skeleton. Stop when the skeleton stabilizes, another pass adds little, or the user has enough to test.