Orthogonal Insight Protocol Toolkit

Generate non-obvious solutions by changing the physics of your problem.

What Is This?

The Orthogonal Insight Protocol escapes "median" AI answers by:

1. Build a world with impossible physics (from a random seed word)
2. Solve the problem inside that world (with a fresh, blind agent)
3. Extract the mechanism — the abstract principle that made it work
4. Bridge to reality — implement the mechanism in the real world

Key insight: Temperature gives you different words. Seeds give you different logic.

Quick Start (Claude Code)

1. Clone this repo and open in Claude Code:

   git clone https://github.com/emergent-wisdom/orthogonal-insight-toolkit
   cd orthogonal-insight-toolkit
   claude

2. Describe your problem:

   How can we reduce loneliness in cities?
   

Claude Code will spawn separate agents for each phase (world builder, solver, extractor) to ensure blind operation.

Installation

Just clone and open - the CLAUDE.md file tells Claude how to run the protocol.

How It Works

The Problem with Temperature

When you crank up AI temperature, you get lexical diversity — different words for similar ideas. The model is still "standing in the same spot."

The Solution: Semantic Seeds

When you seed with impossible world constraints ("everything is porous", "violence mirrors back on the actor"), you get semantic diversity — genuinely different problem frames.

Why Fresh Agents Matter

Each phase uses a separate agent with no shared context:

• World Builder knows only the seed (not the problem)
• Solver knows only the world rules (not the seed)
• Extractor knows only the solution (not the experiment)

This prevents contamination. The blind architecture forces genuine divergence.

Examples

Loneliness × POROUS

Problem: Epidemic of loneliness Seed: POROUS (nothing maintains boundaries)

Standard approach: More social programs, dating apps, community centers

OIP insight: In a porous world, loneliness itself is the contagion spreading through connections. It's not a connection deficit — it's a transmission problem.

Bridge to reality: Cacioppo's Framingham study (2009) found loneliness spreads through social networks like a contagion. Train "loneliness composters" who absorb and transform it.

Antibiotics × MIRRORED

Problem: Antibiotic resistance Seed: MIRRORED (all actions reflect back on the actor)

Standard approach: Use antibiotics more carefully

OIP insight: In a mirrored world, you can't attack pathogens at all. You must make beneficial bacteria outcompete harmful ones.

Bridge to reality: Fecal microbiota transplants, quorum sensing disruption, environmental optimization over direct intervention.

When to Use This

• You keep getting the same obvious answers
• The problem feels "stuck" in a particular framing
• You want to challenge assumptions
• The value of a breakthrough is 100x a standard improvement

When NOT to Use This

• Simple/routine tasks
• The obvious solution is correct
• You need a quick answer

Don't use the Orthogonal Insight Protocol to build a toaster. Use it to invent the microwave.

Credits

Based on "Orthogonal Insight Protocol" by Henrik Westerberg.

License

MIT