magic-wand-number — quality + safety report

---
name: magic-wand-number
description: Apply Elon Musk's Magic Wand Number — the hypothetical cost or performance floor if you could rearrange atoms for free — to set ambition, attack cost reduction, calibrate targets, and expose how inefficient your current process really is. Use this skill when the user is goal-setting, debating performance targets, capacity planning, calibrating ambition, asking "what should our target be?", attacking unit cost, justifying a budget, accepting a vendor quote, scoping latency or throughput, or doing any "is this number actually good?" sanity check. Use it whenever the user reasons by analogy to historical or competitor numbers ("rockets have always cost X, so ours will too", "industry standard is Y") instead of from physical or material limits. Also trigger on phrases like "first principles", "bill of materials", "BOM cost", "idiot index", "what's possible", "how cheap could this get", "how fast could this be". Trigger eagerly even when the user does not name Musk or the framework.
stacks_with:
  - thinking-in-limits
  - idiot-index
---

# The Magic Wand Number

> "If you have them stacked on the floor and could wave a magic wand to create the rocket, what would the cost of the rocket be? We imagine the cost of rearranging the atoms was zero. That's going to set the floor of the cost of the rocket. I call this the 'magic wand number,' the hypothetical best-case scenario."
> — Elon Musk, *The Book of Elon* (Chapter: First-Principles Thinking)

## What this skill captures

The Magic Wand Number is the cost (or latency, mass, energy, headcount, time) of a thing if assembly were free. You list the raw inputs, price them at their commodity market, sum, and that is the physical floor. Everything above that floor is overhead — manufacturing, process, design complexity, organizational drag — and is therefore attackable. Reasoning by analogy ("rockets have always been expensive, therefore ours will be") hides this floor. The Magic Wand Number exposes it. For SpaceX rockets, the floor turned out to be 1–2% of prevailing rocket cost. That gap was the opportunity.

This skill forces the user to compute that floor before accepting any quote, target, or budget.

## When to use this skill

Trigger when the user is:
- Setting a cost, price, or performance target ("what should our unit cost be?")
- Accepting a vendor or internal quote without challenge
- Doing capacity, latency, or throughput planning
- Calibrating ambition on a roadmap or OKR
- Reasoning by analogy ("competitors charge X", "industry benchmark is Y", "we've always done it in Z weeks")
- Saying something is "as cheap/fast as it can be" without showing the floor
- Asking "is this a good number?"

Trigger eagerly. Most goal-setting conversations are reasoning by analogy in disguise.

## The how-to

1. **Name the artifact and the metric.** What thing, what dimension? (Dollars per kWh, seconds per request, grams per unit, minutes per onboarding.) Be specific. A vague artifact gives a vague floor.

2. **Decompose into irreducible inputs.** Bill of materials for a physical thing; CPU cycles, network round-trips, and bytes for software; human-minutes of irreducible decision for a process.
   > "A rocket is made from aluminum, titanium, copper, and carbon fiber. Break it down further and ask, 'How much of each material is used?'"
   > — *The Book of Elon* (First-Principles Thinking)

3. **Price each input at its commodity floor.** London Metal Exchange, spot cloud pricing, minimum-wage human time, Shannon limit, speed of light. Not vendor price. Not "what we paid last time." The market or physical floor.

4. **Sum to the Magic Wand Number.** This is the floor: the hypothetical best-case scenario assuming rearranging atoms is free.

5. **Compute the Idiot Index = current cost / magic wand number.**
   > "A component that costs $1,000 when the aluminum it was made of costs only ten dollars likely has a design that is too complex or an inefficient manufacturing process. If the ratio is high, you're an idiot."
   > — *The Book of Elon*

6. **Declare the gap and the target.** State the floor explicitly. Set the target as a specific fraction of the gap to close in this cycle. Refuse any target that is set as a small percentage delta off today's number.

7. **Reframe the work.** The remaining question is no longer "can we afford it?" but "how do we get the atoms in the right shape more efficiently?"
   > "Our challenge was to figure out how to get the atoms in the right shape more efficiently."
   > — *The Book of Elon*

## Common failure modes

- **Pricing inputs at retail or vendor cost.** That smuggles the inefficiency you are trying to expose back into the floor. Use the commodity / physical limit.
- **Decomposing only one level.** "A server" is not an input. CPU-seconds, RAM-GB-hours, and egress-GB are.
- **Treating the Magic Wand Number as a target.** It is the floor, not the goal. The goal is a chosen fraction of the gap you commit to closing.
- **Skipping the Idiot Index.** Without the ratio, the gap stays abstract. A 50x ratio is a different conversation than a 1.5x ratio.
- **Quietly adding back "necessary overhead" before showing the raw floor.** Show the naked number first. Then justify each addition line by line, or delete it.
- **Reasoning by analogy mid-exercise.** "But competitor X also charges this much" is the exact failure mode the skill exists to kill.

## When NOT to use this skill

- The constraint is regulatory, contractual, or social, not physical (e.g., licensing fees, union rates by contract). Compute the floor anyway, but the gap may not be attackable in the current cycle.
- The artifact is truly novel and has no decomposable inputs yet (very early R&D where the bill of materials itself is the unknown). Use first-principles exploration instead.
- The decision is qualitative (hire vs. not hire a specific person, pick a brand name). The Magic Wand Number is for quantitative floors.

## Source

The Book of Elon by Eric Jorgenson (2026, Scribe Media). Chapter: "First-Principles Thinking" (Magic Wand Number subsection).