feynman-science

name: feynman-science description: Feynman Scientific Method. Use when relevant to this domain. domain: research

Feynman Scientific Method

Core Philosophy

"The first principle is that you must not fool yourself — and you are the easiest person to fool."

The Feynman Mantra:

1. Simplify - If you can't explain it simply, you don't understand it
2. Verify - Don't trust anything without evidence
3. Question - Especially authority and common knowledge
4. Iterate - Keep testing, keep learning
5. Connect - Find analogies that reveal truth

The Feynman Problem-Solving Method

Explain the concept as if teaching a child. Identify gaps in your explanation.

Step 1: Write Down the Problem

"If you're stuck, write down the problem clearly. The act of writing forces clarity."

• Define exactly what you're trying to solve
• State constraints and knowns
• Identify what's unknown

Step 2: Make Your Best Guess

"Guess and check is a valid method."

• Form a hypothesis (even a wild one)
• Don't wait for perfect information
• Better to try and fail than not try at all

Step 3: Check Your Guess

"It doesn't matter how beautiful your theory is. If it doesn't agree with experiment, it's wrong."

• Design a test
• Run the test
• Record results honestly
• Accept when wrong

Step 4: When Stuck, Simplify

"Look at the problem from a different angle."

• Break into smaller problems
• Try simpler versions
• Draw a diagram
• Explain to someone (or a rubber duck)

Step 5: Connect to Known

"Find an analogy. Your problem is like what?"

• Connect to something you understand
• Use that understanding
• Transfer the insight

Scientific Integrity

Never fake data, cherry-pick results, or overstate confidence.

The Feynman Test:

Before publishing/presenting:
├── Is this reproducible?
├── Have I eliminated bias?
├── Am I hiding inconvenient data?
├── Would I be embarrassed if proven wrong?
└── Can a smart person disagree?

Known Unknowns vs Unknown Unknowns:

Learning Framework

Build understanding from first principles, then connect to existing knowledge.

The Feynman Technique:

1. Choose a concept to learn
2. Teach it to a child (write out simply)
3. Identify gaps - where did you struggle?
4. Review and simplify - fix gaps, try again

The Curiosity-Driven Path:

Interest → Question → Research → Hypothesis → Test → Learn → New Questions
                                                  ↓
                                              Share!

Problem-Solving Heuristics

Techniques for breaking down hard problems into tractable pieces.

The 5-Why Method:

Problem: Why X? Answer: Because Y

Problem: Why Y? Answer: Because Z

...until root cause

First Principles Physics:

Question any assumption:
├── "Why is this true?"
├── "What if the opposite were true?"
├── "What physics applies here?"
├── "What's the limit of this approach?"
└── "Am I confusing correlation with causation?"

The Devil's Advocate:

• Before concluding, find the flaw
• Ask: "What's the best argument against this?"
• Test that argument
• Update beliefs based on result

Quotes to Live By

"I was taught that the harder you try, the farther ahead you are."

"Science is a way of trying not to fool yourself."

"I would rather have questions that can't be answered than answers that can't be questioned."

"The world is much more interesting than any one discipline."

When NOT to Use

• When the research requires access to proprietary databases or paywalled sources
• When findings will be used for financial decisions requiring licensed advisor review
• When the task is too trivial to warrant this skill
• When a more appropriate skill exists

Common Rationalizations

Red Flags

• Research relies on a single unverified source
• Agent presents speculation as confirmed findings
• Watch for shortcuts and skipped steps

Verification

After completing this skill, confirm:

• Findings are verified across multiple independent sources
• Research methodology is documented and reproducible
• All required outputs generated
• Success criteria met

Related Skills

• brainstorming - Problem exploration
• systematic-debugging - Finding errors
• mckinsey-research - Strategic research
• continuous-learning - Improving over time