Meta First Principles Agent Skill

First Principles Thinking

Framework

IRON LAW: Assumptions Are Not Facts

Most "constraints" are assumed, not physical. "We can't do X" usually means
"We haven't seen anyone do X" or "The current method doesn't allow X."
First principles thinking distinguishes between:
- Physical laws (actual constraints: gravity, thermodynamics)
- Conventions (assumed constraints: "this is how it's done")

The Method

1. Identify the assumption: What do "everyone knows" or "everyone does"?

2. Break it down to fundamentals: What are the basic, irreducible facts?

• Techniques: "Five Whys" (ask why 5 times), cost decomposition, physics-level analysis

3. Reason up from fundamentals: Given only the base facts, what's the best solution?

Example: Elon Musk on Battery Costs (2010s)

• Convention: "Batteries cost $600/kWh. Electric cars will always be too expensive."
• First principles: What are batteries made of? Cobalt, nickel, lithium, carbon, metals, polymers. What do these raw materials cost? ~$80/kWh at commodity prices.
• Reasoning up: If we buy materials and build batteries ourselves, we can potentially reduce costs to near $80/kWh. The $520 gap is manufacturing inefficiency and supply chain markup, not physics.

Five Whys for Assumption Drilling

• "Why is our product priced at $100?" — Because our COGS is $60 and we need 40% margin.
• "Why is COGS $60?" — Because we use supplier X for component Y.
• "Why do we use supplier X?" — Because we've always used them.
• "Why haven't we looked at alternatives?" — Because the procurement team hasn't been asked to.
• "Why not?" — There's no real reason. It's just convention.

→ Fundamental constraint found: the COGS includes an unexamined supplier relationship, not a physical constraint.

Cost Decomposition Template

For any "it costs too much" problem:

• List every cost component
• For each, identify: Is this cost from physics (materials, energy) or from convention (markup, inefficiency, legacy process)?
• Question every convention-based cost

Output Format

# First Principles Analysis: {Problem}

## The Assumption
{What "everyone knows" or the conventional constraint}

## Decomposition
| Component | Cost/Fact | Physics or Convention? |
|-----------|----------|----------------------|
| {element} | {value} | Physics / Convention |

## Fundamental Truths
{What remains after stripping away assumptions}

## Reasoning Up
{Given only fundamental truths, what's the best approach?}

## Proposed Solution
{What changes if we ignore conventions and reason from fundamentals}

Gotchas

• Don't reinvent the wheel for everything: First principles is expensive (time, effort). Use it for high-stakes problems where conventional approaches clearly fail. For routine decisions, best practices and heuristics are fine.
• Physics constraints are real: Don't confuse first principles thinking with ignoring physical reality. You can't reason your way past thermodynamics.
• "Convention" has value: Conventions encode collective experience. Dismissing all conventions as "assumptions" is arrogance, not first principles thinking. The point is to EXAMINE conventions, not automatically reject them.
• Execution matters: A first-principles insight is worthless without execution capability. SpaceX didn't just THINK about cheap rockets — they BUILT a vertically integrated manufacturing operation.

References

• For Socratic questioning method (complementary), see the hum-socratic skill