The Impossible Rocket Company — Why Reasoning From the Ground Up Beats Reasoning by Analogy

Musk and the Impossible Rocket Company

In 2002, Elon Musk was told it was impossible to build a private rocket company. The established aerospace industry had a clean consensus on the matter. The engineering was too complex, the capital requirements were too enormous, and the regulatory environment was too hostile. Even sympathetic advisors told him to find a different use for his PayPal money.

So Musk started asking different questions, and this is the part I want you to notice. Not “can a private company launch rockets.” Instead: what are rockets actually made of? What do the materials actually cost? What’s the real minimum cost per kilogram to reach orbit if you build from scratch with modern manufacturing? He traced every assumption back to physics, not to industry convention. SpaceX reached orbit in 2008. By 2023 they’d reduced the cost of reaching orbit by a factor of roughly twenty.

That’s first principles thinking in operation, and I want you to think of it that way from here on out. Not as a philosophy seminar topic. As a weapon you’re going to learn to use on your own stuck problems.

First principles thinking is the practice of decomposing a problem down to its fundamental, irreducible truths — the bedrock facts that can’t be derived from anything simpler. Then you rebuild your understanding from that foundation, rather than from inherited assumptions, industry conventions, or analogical reasoning. It’s how Aristotle defined the philosophical foundations of scientific reasoning. It’s how Richard Feynman approached every physics problem he ever encountered. It’s what separates the people who genuinely solve problems from the people who just manage their inheritance of other people’s solutions.

Most people, most of the time, reason by analogy, and you probably do too more often than you’d like to admit. You look at what already exists and ask: how can I modify that? Tim Urban of Wait But Why describes this as “the cook’s approach” versus “the physicist’s approach.” The cook follows a recipe. The physicist understands the chemistry well enough to invent new recipes, or to recognize when the recipe is wrong at a fundamental level. Analogy reasoning is fast, and it’s often adequate for your routine problems. It catastrophically fails you on novel problems, because the analogy you’re using was calibrated for a different situation entirely.

This episode is the Decomposition Protocol — a systematic framework for breaking any problem you have down to first principles and rebuilding it from the ground up. We’ll cover what first principles actually are, why your analogical reasoning fails you, how Feynman and Munger and Parrish operationalize this in practice, and exactly how you apply the protocol to the specific problems stalling you right now.

The History of the Idea and Why It Matters to You

Aristotle introduced the concept in the Posterior Analytics. He defined a “first principle” as “the first basis from which a thing is known” — a foundational truth that cannot be derived from any prior truth. He was making a point about the structure of knowledge you should sit with. You can only reason backwards so far before you hit bedrock. Everything else you know is built on top of that bedrock, either directly or through chains of inference. The quality of your own reasoning depends entirely on whether you’re building on solid bedrock or on assumptions that someone else built on assumptions.

In physics, this became the standard method, and it’s the standard you should be holding yourself to. You don’t accept a principle because an authority said it. You trace it back to experimental evidence and mathematical derivation. If you can’t trace it back, it’s a hypothesis for you, not a principle.

Richard Feynman made this viscerally concrete. In his famous “Cargo Cult Science” lecture at Caltech in 1974, he argued that most of what passes for scientific thinking is actually the ritualistic imitation of science — using the words and forms of scientific reasoning without the substance. You can look like you’re doing rigorous thinking while actually just recycling received wisdom in a structured format. Real first principles reasoning requires you to be willing to discard everything you inherited if it doesn’t survive contact with the actual evidence.

Feynman’s famous approach to learning anything applies directly to you: read the textbook, close the textbook, try to derive everything from scratch. If you can’t derive it from scratch, you don’t actually understand it. You’ve just memorized it. Memorization can be disrupted in you. Understanding is load-bearing.

Shane Parrish at Farnam Street has spent years translating this method into business and personal decision-making for people like you. His mental models framework draws heavily on Charlie Munger’s “latticework of models.” It’s essentially an argument that the way you reason from first principles in complex domains is to build a rich library of foundational concepts from multiple disciplines, and use them to triangulate on your novel problems. You’re not reasoning from zero every time. You’re reasoning from a carefully vetted set of bedrock principles drawn from physics, psychology, economics, biology, and mathematics.

Charlie Munger, Warren Buffett’s partner at Berkshire Hathaway, articulated this in his famous 1994 USC speech.

“I constantly see people rise in life who are not the smartest — sometimes not even the most diligent — but they are learning machines. They go to bed every night a little wiser than they were when they got up. And boy, does that help — particularly when you have a long run ahead of you.” — Charlie Munger, USC Commencement Address, 1994.

The “learning machine” model is first principles reasoning applied to your own self-improvement. It means constantly updating your foundational models, rather than operating on the assumptions you built in your twenties and never revisited.

Why Analogical Reasoning Fails You When It Matters Most

Before we get into the protocol, you need to understand the failure mode it’s designed to prevent. Analogical reasoning — reasoning by comparison to similar cases — is your default mode of thinking, and for good reason. It’s fast, it draws on your experience, and it works well in stable, familiar domains. The problem is that it fails you precisely at the moments when you need reasoning most: novel situations, disrupted industries, unprecedented challenges, complex systems.

Here’s the mechanism of failure. When you reason by analogy, you’re importing not just the useful structural features of the comparison case. You’re importing all the assumptions baked into it, including the ones that are wrong, outdated, or irrelevant to your actual situation. The analogy comes as a package. You don’t get to select only the true parts and discard the rest.

Let me give you a concrete example. When digital photography emerged in the 1990s, Kodak’s response was shaped heavily by analogical reasoning. This is a new format, like the transition from glass plates to film. We own the chemical processing infrastructure. The new format will plug into our existing infrastructure. We’ll control the processing. This analogy was seductive because it was historically accurate — Kodak had successfully made the glass-plate-to-film transition exactly that way before. But the analogy failed at the fundamental level, because digital photography does not require processing. The entire infrastructure value proposition — the pharmacies, the processing labs, the print facilities — was never transferring to the new medium. The analogy dragged in an assumption that was not only wrong but catastrophically wrong. Kodak invented the digital camera in 1975 and killed it internally. They filed for bankruptcy in 2012.

Here’s a third example, a composite I’ll call Tom Reeves — built from a pattern common among consultants who go independent, not one specific person. He’d spent eight years as a partner at a management consulting firm before launching an advisory practice of his own. Tom’s mental model for client acquisition was entirely analogical. He’d built his book of business at the consulting firm through internal referrals, reputation within the partnership, and assigned accounts. He tried to apply the same pattern to his independent practice — build a reputation, network with other advisors, wait for referrals. Three years later he had twelve clients. His analogy was structurally accurate, but it ignored a first-principles difference you need to notice: at the consulting firm, the institution did the primary selling. As an independent, he was the institution, and the institution had no reputation yet and no existing trust network. The analogy mapped the relationship-building correctly but missed the foundational difference in institutional credibility. Once he identified that first-principles difference, he completely redesigned his go-to-market approach.

What First Principles Actually Are, and Are Not, For You

Here’s where people get confused, so let me be precise with you. A first principle is not just a foundational assumption. It’s a foundational truth — something that survives your rigorous interrogation and cannot be reduced any further. The distinction matters, because you’ll be tempted to mistake conventional wisdom for a first principle. You’ll trace your reasoning back to “this is how it’s always been done” or “this is industry standard” and call that a first principle. It isn’t. “This is how it’s always been done” is an observation about past behavior for you, not a truth about what must be done.

The true first principles available to you in business tend to be drawn from physics — what are the actual energy and material requirements? From economics — what are the actual incentive structures? From human psychology — what do people actually want, not what you assume they want? And from mathematics — what are the actual numbers? These are domains where you can find bedrock truths that don’t shift based on convention.

Shane Parrish distinguishes between “what I believe” and “what I can verify.” Your first principles work requires moving your load-bearing reasoning into the “can verify” column. Everything sitting in your “what I believe” column is a hypothesis, not a foundation.

Feynman had a practical test for whether you actually understand something versus whether you’ve merely memorized it. He called it the Feynman Technique: explain the concept in plain language to a twelve-year-old. Every time you hit jargon, you’ve found a gap in your own understanding — a place where you’re using a word as a substitute for the understanding you don’t actually have. Go back to the source material and fill that gap. Repeat until you can give a clean, plain-language explanation. That explanation is built on first principles. Everything before that was imitation.

The Decomposition Protocol: Six Steps for You

Here’s the practical framework I promised you. Six steps for taking any problem you have, stripping it to its foundations, and rebuilding a solution from the ground up.

1. Define the problem precisely. Most people work on vague problems, which means they find vague solutions. Your first principles work begins with a crisp, specific problem statement. Not “my business isn’t growing fast enough” but “my customer acquisition cost is $340 and my lifetime value is $280, which means I’m destroying value with every customer I acquire. The specific problem is: either reduce CAC below $280 or increase LTV above $340.” The more precise your problem statement, the more tractable the decomposition becomes.
2. List every assumption embedded in how you’re currently thinking about the problem. Write them all down. Every “of course” and “obviously” and “everyone knows” — these are your assumptions. Get them visible and explicit to yourself. You’re looking for places where you’re accepting inherited thinking as bedrock truth. Pay particular attention to your constraints: “I can’t do X because” — that “because” almost always contains an assumption rather than a first principle.
3. Challenge every assumption you wrote down. For each one, ask yourself: is this actually true? How do I know? What would the world look like if this were false? What evidence would I need to see to update this belief? This is where Feynman’s warning about not fooling yourself applies most directly to you. The goal is to separate what’s actually true from what you’ve merely assumed to be true. The distinction is load-bearing.
4. Identify the actual bedrock truths. After stripping away the assumptions, what remains for you? What are the physical facts, the mathematical relationships, the empirically verified psychological truths, the economic incentive structures? These are your first principles — the things you can build on because they’re real, not inherited. Write them down explicitly.
5. Rebuild from bedrock. Starting from your first principles only — explicitly excluding the assumptions you identified — build a new solution to your problem. What would this look like if it were designed from scratch, with no legacy constraints, no industry conventions, no “this is how it’s always been done”? This is where genuinely novel solutions emerge for you. You’re not constrained by the history of the domain. You’re constrained only by what’s actually true.
6. Test the rebuild against reality. Your rebuilt solution is a hypothesis. It has to be tested. First principles work doesn’t produce certainty for you. It produces better hypotheses. The difference from your ordinary hypotheses is that your reasoning is load-bearing, so when the test fails, you understand precisely which assumption or first principle needs updating. Your feedback loop gets tighter and more productive every time you run it.

Feynman’s Version: The Physics of Problem-Solving

Richard Feynman won the Nobel Prize in Physics in 1965 for his work on quantum electrodynamics — his path integral formulation of quantum mechanics. But his greatest intellectual contribution, the one most useful to you, was arguably his approach to knowing things at all.

Feynman explicitly rejected the idea that you could learn by reading about learning. His approach was always to work the problem. Derive the result yourself. When he was studying physics in college, he systematically worked through every major result in every textbook from first principles rather than accepting the proof as given. When he worked on the Manhattan Project at Los Alamos, he earned a reputation for solving problems that senior physicists found intractable. Often that was because he’d strip a problem down to its physics while everyone else was stuck in inherited mathematical formalisms.

His diagnostic question — the one he applied to every claim he encountered, and the one you should start applying too — was simple: what experiment would prove this wrong? This isn’t skepticism for its own sake. It’s the test that separates claims that are load-bearing from claims that are decorative. A claim that cannot, even in principle, be proven wrong is not a first principle. It’s an unfalsifiable belief, and unfalsifiable beliefs don’t belong anywhere in your foundation.

The Feynman Technique, which he never formally named but which is documented extensively in his teaching and in James Gleick’s biography Genius, has four steps for you to use. First, choose a concept you want to understand. Second, explain it in simple language as if teaching a child. Third, identify the gaps where your explanation breaks down or requires jargon. Fourth, go back to the source and study until you can fill the gap with plain language. Repeat. The cycle continues for you until the explanation is airtight in plain language. That’s understanding. Everything before that is performance.

“The first principle is that you must not fool yourself — and you are the easiest person to fool.” — Richard Feynman, Caltech Commencement Address, 1974.

Munger’s Latticework: First Principles Across Disciplines

1. The central limit theorem, from statistics.
2. Comparative advantage, from economics.
3. Natural selection, from evolutionary biology.
4. Compound interest, from mathematics.
5. Confirmation bias, from psychology.
6. The principle of inversion, from logic.

Charlie Munger’s approach to first principles is different from Feynman’s in one important respect that matters for how you should use it. Munger worked in complex, multi-domain problems where no single discipline provides all the relevant first principles. Business strategy, investment analysis, organizational behavior, macroeconomics — these problems involve interacting systems from multiple domains, and you need first principles from multiple disciplines to triangulate correctly.

Munger’s solution was what he called the latticework of mental models — a deliberately cultivated library of the most powerful and reliable principles from multiple disciplines. Not a shallow survey of every field for you to skim, but a deep mastery of the most load-bearing ideas in each. Here’s a starting set worth building into your own latticework.

The value of the latticework is that it lets you triangulate. When a business problem presents itself to you, you’re not reasoning from analogy. You’re running the problem through multiple first-principle lenses and looking for where the answers converge. Convergence from independent perspectives is the closest thing to certainty available to you in complex problems.

Munger’s most famous first principle is inversion. He borrowed it from the mathematician Carl Jacobi, who famously advised “invert, always invert.” Before you try to figure out how to achieve a goal, figure out what would guarantee your failure. If you want to build a successful business, first make a comprehensive list of everything that would guarantee the business fails. Then avoid those things. The inversion surfaces risks and failure modes that your forward-directed thinking misses, because your mind is optimistically biased when you’re approaching goals you want to achieve.

Shane Parrish at Farnam Street built an entire educational platform on extending Munger’s latticework model for people like you. His Great Mental Models series systematically documents the most powerful and transferable principles from physics, biology, systems thinking, mathematics, human judgment, and military strategy. The underlying argument — which I find entirely persuasive, and I want you to hear it directly — is that most smart people are domain-confined. They reason expertly within their own field and by analogy or intuition outside it. Building a true latticework means you have load-bearing first principles available across domains, which is the difference between reasoning with a full toolkit and reasoning with a single instrument.

Tim Urban’s Framework: Chefs and Cooks

Tim Urban of Wait But Why wrote one of the most accessible explanations of the cook-versus-chef distinction, and it maps directly onto your analogical-versus-first-principles reasoning. The cook follows recipes. The chef understands the underlying chemistry and flavor science well enough to create new recipes from scratch, and to improvise successfully when ingredients are missing.

Urban’s point is that you, like most people, are probably a cook in your own life more often than you’d admit. You follow inherited recipes — career paths, relationship structures, financial strategies, life sequences — that were written by someone else for someone else’s circumstances. The cook’s version of success is following the recipe correctly. The chef’s version is understanding why the recipe works, which lets you adapt it or replace it when your circumstances change.

This isn’t a criticism of convention. Some recipes are excellent, and you shouldn’t throw them out reflexively. The question is whether you’re following them because you’ve verified they’re right for your situation, or because you’ve never thought to question them. A chef can choose to follow a standard recipe when it’s optimal. A cook has no choice — they only have the recipe.

Urban traces Musk’s SpaceX thinking specifically.

“Someone could tell Elon that the reason we can’t yet have fully reusable rockets is that the technology doesn’t yet allow for it. Then Elon would think to himself, ‘Is that actually true? Let me find out.’ And he’d read everything there is to read about rocket construction and conclude that if you build the rocket right, reusability is entirely achievable with current technology.” — Tim Urban, Wait But Why.

The existing industry consensus wasn’t a first principle. It was a cook’s reading of the recipe that had never been verified against actual physics.

Here’s a fourth composite for you — I’ll call her Rachel Torres, built from a pattern common among agency owners who hit a growth ceiling. She ran a digital marketing agency for eleven years. By year eight, she was trapped in a $2 million agency that couldn’t grow past its founder’s own billable hours. The standard advice in her industry was to hire a sales team and build a delivery team to replace herself. She tried this twice and failed both times. She was reasoning from the industry recipe. When she finally applied first principles, she traced the constraint back to fundamentals. Her business model required custom delivery for every client, which meant her expertise couldn’t be systematically replicated, which meant every new team member required extensive training on judgment, which meant linear growth was impossible without degrading quality. The real problem wasn’t sales or delivery. It was a productization problem. The model needed to change before the headcount could scale. She rebuilt around a productized service and crossed $5 million in year twelve.

When First Principles Thinking Is Most Valuable to You

1. You’re stuck. If you’ve tried the obvious approaches and they’ve failed, the most likely explanation is that you’re working from a false assumption. First principles work finds the false assumption for you.
2. The domain is changing rapidly. When your environment shifts, the analogies calibrated on the old environment fail you. First principles let you build from bedrock that doesn’t change even when the surface-level landscape does.
3. You’re entering a new field. When you’re new to a domain, you have no reliable analogies from your own experience to lean on. First principles work lets you build a foundation rather than borrow someone else’s potentially flawed one.
4. The consensus seems wrong to you. When everyone in an industry agrees on something that strikes you as odd, first principles work lets you check whether the consensus is grounded in actual truth or in inherited convention. Sometimes the consensus is wrong for reasons visible only from the bedrock level.
5. High stakes, irreversible decisions. When the cost of being wrong is large and the decision can’t easily be undone, your investment in first principles reasoning pays for itself.

First principles thinking is not always the right tool for you. For routine, well-understood problems in stable domains, analogical reasoning is faster and often adequate. You don’t need to derive thermodynamics from quantum mechanics every time you want to heat your house. The time and cognitive cost of first principles decomposition is only justified for you when the stakes are high and the problem is novel, ambiguous, or recurring without resolution.

Here are the specific conditions where first principles thinking delivers disproportionate value to you.

The Practical Discipline: Making It a Habit

The challenge with first principles thinking is that it requires active cognitive effort from you in situations where your brain is actively trying to shortcut to the analogical answer. Kahneman’s System 1 is pattern-matching and analogizing in you continuously. First principles work requires you to stop that process and run a deliberate decomposition instead. Your brain resists this, because it’s effortful.

Feynman’s practice of working through problems from scratch rather than accepting inherited derivations was a daily discipline for him, not a special-occasion technique. He did it because it kept his fundamental understanding sharp and current. When you stop deriving and start accepting, your own understanding fossilizes. The derivations you stopped verifying become the assumptions that destroy careers decades later.

Munger’s approach to building the latticework is a reading discipline you can copy directly. He read across disciplines constantly — not for information, but for first principles. Every time he encountered a concept that struck him as a fundamental truth about how systems work, he added it to his mental model library and practiced applying it to problems outside its original domain. The latticework isn’t built in a weekend for you either. It’s built over decades of deliberate, cross-disciplinary reading and application.

Parrish’s practical recommendation, and one you can start using today, is to maintain a decision journal. For every significant decision you make, write down the problem, the assumptions you identified and challenged, the first principles you’re building on, and the conclusion. Review it quarterly. You’ll find the assumptions that keep recurring for you, the ones you keep failing to challenge, and the first principles you keep mis-applying. The journal is a calibration tool for your own reasoning process, not just a record of decisions you’ve already made.

This connects directly to the episode on cognitive biases — specifically the Decision Audit Protocol. First principles thinking is the upstream process for you. The Decision Audit catches the downstream biases that distort what you do with your first principles once you’ve actually found them.

For the related skill of protecting your first principles reasoning from manipulation by others, there’s the episode on Dark Psychology. Manipulators exploit your analogical reasoning and the assumptions embedded in social conventions. First principles thinking is one of your most reliable defenses against that.

And the application of first principles to questions of exit and continuation — when it’s rational for you to stop, when it’s rational for you to continue — is explored in the episode on the Sunk Cost Fallacy. Stripping away the psychological weight of your past investment and reasoning from first principles about your own future is one of the hardest and most valuable skills in this entire framework.

The Cost of Refusing This Work

Tom Reeves, the consultant, spent three years running a broken business model because he couldn’t see past his own analogy. Rachel Torres spent three years trying to solve a productization problem with a hiring solution. And in a story you’ve probably heard elsewhere on this show, a founder we’ve called Marcus Chen spent six months burning runway on a startup that had already failed by the metrics he himself had set. None of these are stories about stupid people. They’re stories about intelligent people reasoning by analogy in situations that required first principles instead.

The common thread for you to notice: each of them had access to the information they needed to identify the foundational flaw. The information was there the whole time. The problem wasn’t a lack of data. It was a lack of framework for questioning the assumptions that the data was being filtered through. Confirmation bias, anchoring, the availability heuristic — all the biases covered elsewhere on this show — run on the surface layer of your reasoning. First principles work operates at a deeper layer than any of that. It’s not just about correcting for biases in how you evaluate information. It’s about questioning whether the fundamental frame you’re using to think about your problem is correct in the first place.

Aristotle’s insight, twenty-three centuries old, is still the right place for you to land on this. You can only reason as well as your foundations allow. Inherited assumptions corrode your foundations silently, without announcing themselves. First principles work is the maintenance protocol that keeps them sound for you.

Know your foundations. Question them. Build on what’s real.

The Feynman Technique in Practice: A Step-by-Step Walkthrough

Because the Feynman Technique is one of the most powerful single tools in this arsenal, it deserves a practical walkthrough from me rather than just a description. The technique has four steps, and each one does specific cognitive work on you.

Step one: choose a concept. Not a domain — a specific concept. Not “strategy” but “competitive moat in a two-sided marketplace.” Not “leadership” but “how psychological safety affects team risk-taking in high-consequence environments.” The more specific your concept, the more useful the technique becomes. Vague concepts let you hide vague understanding behind vague language. If you can’t narrow your concept to a single sentence that identifies the precise mechanism or relationship you need to understand, you haven’t defined your concept yet.

Step two: explain it in plain language as if you’re teaching a twelve-year-old. Write it out completely. Not just think about it — write it out. The act of writing forces complete articulation from you. When you’re only thinking, you can skip over the parts you don’t fully understand, because your brain pattern-matches across the gap without you noticing. When you write it out, the gap appears as an incomplete sentence, a word you’re using as a placeholder, or a logical jump your reader couldn’t follow. The writing surface is where your understanding becomes visible, and where the absence of understanding becomes undeniable to you.

Step three: identify every gap and every piece of jargon in your own explanation. Jargon isn’t knowledge for you. It’s a container you haven’t opened. If your explanation includes words like “synergy,” “value creation,” “alignment,” “disruption,” or any technical term you haven’t fully explained, you’ve found a gap. Go back to the source material and replace each piece of jargon with the specific mechanism it’s supposed to describe. If you can’t do that — if the jargon doesn’t decompose into a describable mechanism — the jargon is a placeholder for understanding you don’t actually have, and the claim you built on it is built on nothing.

Step four: simplify and refine. After filling the gaps, rewrite your explanation from scratch. Your goal is an account that’s both complete and clear — one that doesn’t rely on jargon, doesn’t skip steps, and could genuinely be followed by someone with no prior domain knowledge. This final explanation is your first principle. Everything built on top of it is derivation. Every derivation you can’t trace back to this foundation is an assumption you’ve been treating as knowledge.

Tom Reeves used this technique when he was trying to understand why his client acquisition model wasn’t working. He tried to explain in plain language exactly how a new potential client was supposed to find him, decide he was credible, and choose to engage him. When he wrote it out, his explanation for step two — how does the client decide I’m credible — reduced to the single word reputation. He tried to explain reputation without jargon next. He realized it meant what other people say about him to potential clients. When he traced that mechanism further, he realized it depended entirely on referrals. Referrals required existing clients who were actively talking about him. That required either a larger client base than he currently had, or a completely different credibility mechanism altogether. The word reputation had been hiding the complete absence of a functional client acquisition mechanism the entire time. The Feynman Technique surfaced this for him in twenty minutes. Three years of stagnation had not.

Munger’s Inversion: Your First Principle for Avoiding Failure

Of all the first principles in Munger’s latticework, inversion is the one most immediately applicable to whatever problem you’re currently stuck on. The mathematician Carl Jacobi’s famous advice — invert, always invert — isn’t a philosophical preference. It’s a problem-solving technique based on a specific insight: many problems that are difficult to solve in the forward direction are trivially easy to solve in reverse, and the reverse solution tells you exactly what to avoid.

Here’s how you apply it. If you want to know how to be happy, first ask what would guarantee your misery, then avoid those things. If you want to know how to build a successful business, first ask what would guarantee its failure, then avoid those things. If you want to know how to be a good leader, first ask what behaviors would guarantee your team loses confidence in you and produces their worst work, then avoid those behaviors.

The reason inversion works better than direct forward-reasoning for you is that your brain is better at identifying concrete failures than at constructing concrete successes. Your successes are often vague to you. Your failures are specific and visualizable. You can picture what happens when a product launch fails because of poor distribution planning. You can picture the exact moment a team disintegrates after a leader dismisses someone’s idea publicly in a meeting. Failures have texture for you. They have sequence. They’re imaginable in ways that abstract success often isn’t. Inverting forces you to get concrete about what you’re trying to avoid, which is almost always more tractable for you than the abstract positive goal you’re trying to achieve.

Practically: before you start any significant new project, spend thirty minutes generating the most comprehensive list you can of the specific actions and omissions that would guarantee it fails. Be specific with yourself. Not poor planning, but launching without validating that the target customer can actually afford the price point at the volumes required for the unit economics to work. Not bad culture, but hiring primarily for technical skills without evaluating how candidates handle disagreement and deliver critical feedback upward. Your list of specific failure modes is your risk map. Avoiding everything on it doesn’t guarantee your success, but it dramatically increases your odds by removing the most common causes of failure from your path before you even start.

Rachel Torres runs an inversion exercise at the beginning of every significant strategic initiative now. She calls it the ways-this-can-go-wrong meeting, and she runs it before any forward planning gets formally presented. The explicit framing — this is how it fails — creates permission to identify problems that forward-planning framing tends to suppress. In her experience, this meeting consistently surfaces two or three significant structural risks that wouldn’t have been identified in the forward-planning process, at a stage where they’re still cheap to address.

Why Domain Experts Are Often the Worst at This

One of the most counterintuitive findings in the research on first principles thinking is that deep domain expertise can actually impede it in you. This seems wrong at first. Shouldn’t the expert be better positioned to identify foundational truths in their own field? The problem is that expertise is built through pattern recognition inside existing frameworks. The more deeply expert you become in a domain, the more completely your thinking gets organized around the existing models and conceptual architecture of that domain. Those models represent the accumulated conventional wisdom, which is mostly, but not always, right. Your facility with the conventional framework makes it much harder for you to see when the framework itself is wrong.

When the conventional wisdom is wrong, experts are often the last to see it, precisely because their cognitive architecture has been so thoroughly shaped by the very conventions they’d need to challenge. Max Planck, the physicist who originated quantum theory, made this observation famously: a new scientific truth doesn’t triumph by convincing its opponents and making them see the light. It triumphs because its opponents eventually die, and a new generation grows up familiar with it. Thomas Kuhn formalized this in The Structure of Scientific Revolutions with his concept of paradigm shifts. Scientific revolutions happen when anomalies accumulate to the point where the dominant paradigm can no longer accommodate them, and a new paradigm has to be constructed from scratch. The people who construct the new paradigm are rarely the deeply invested experts in the old one.

In business, you can see this clearly in every major disruption cycle. Think about the newspaper industry’s response to the internet, the taxi industry’s response to ride-sharing, the hotel industry’s response to home-sharing. In each case, the industry experts understood the technical details better than the disruptors did. But they reasoned from within the existing framework rather than from first principles about what the technology actually made possible. The disruptors were often coming from outside the industry, which meant they had no emotional investment in the existing framework. Their lack of expertise was, counterintuitively, an advantage for them in identifying what was actually possible once you thought about it from scratch. They weren’t carrying the weight of thirty years of industry assumptions.

This doesn’t mean expertise is a liability for you. It means the combination of deep domain knowledge with genuine first principles reasoning capability is rare and extraordinarily powerful. The people who have both can see what’s currently true about a domain and what the foundational principles actually imply — including implications that contradict current practice. They know the domain deeply enough to take it apart at the joints and reassemble it in a configuration the conventional thinkers inside the domain can’t conceive of. Building that combination — deep domain knowledge plus a genuine willingness to challenge the foundations of that domain — is the intellectual project at the heart of this episode. It’s the most valuable cognitive upgrade available to you, if you take it seriously.

Ask yourself, honestly, where you’re the most expert person in the room right now, because that’s exactly where you’re most at risk of this trap, not least at risk. It’s counterintuitive, so sit with it for a second. The domain where you’ve spent the most years is the domain where your pattern-matching is fastest, which means it’s the domain where you’re least likely to notice yourself skipping the decomposition step entirely. You’ll feel certain. Certainty feels like knowledge from the inside, and it feels identical whether it’s earned through first principles or simply inherited through repetition. The test isn’t how confident you feel about your own expertise. It’s whether you can still name the specific evidence underneath your most confident position in your own field, right now, without reaching for “everyone in the industry knows this.” If you can’t, you’ve become a very skilled cook who’s forgotten they were ever supposed to become a chef.

The Decomposition Protocol Applied: A Business Strategy Walkthrough

Let’s walk through this together on a real problem, so you can see the mechanics of the protocol working. A mid-sized software company is losing market share to a competitor that entered two years ago with a lower-priced product. Its leadership team has been debating two options for six months: match the competitor’s price, or add features to justify the existing price premium. Neither option feels right to them. Matching price guts the margin. Adding features takes eighteen months of development and may not even address what customers are actually comparing. The team is stuck.

Step one: define the problem precisely. The problem is not “we are losing market share.” The precise problem is narrower than that. Customer acquisition cost has increased 40% in eighteen months, because prospects who used to convert without a competitive evaluation now compare the company to the low-cost competitor, and they lose 35% of those comparisons. The specific constraint: they can’t match the competitor’s price without eliminating the R&D budget that produces the feature differentiation their existing customers value and that drives retention.

Step two: list every assumption embedded in current thinking. The team assumes the competitive landscape is binary — price or features. They assume customers comparing the two products are evaluating on price alone. They assume the competitor’s pricing is sustainable. They assume the current customer segments are the right segments to defend. They assume the sales process, designed for a no-competition environment, is still the right process for a competitive evaluation situation.

Step three: challenge each assumption in turn. Is the field really binary? Is there a third structural option — different pricing architecture, different segments, partnership models, different distribution channels — that’s never been considered because it doesn’t fit the existing framework? Are customers actually comparing on price, or is there something else driving that 35% loss rate that no one’s checked recently with actual lost prospects? Is the competitor’s pricing model even sustainable, given their reported margins and growth rate? Are the segments being defended the highest-value segments, or simply the most familiar ones? Is the sales process actually optimized for competitive evaluations, or does it assume the prospect has already decided to buy?

Step four: identify bedrock truths. Customers buy on value perception relative to their comparison set, not on absolute capability — that’s a first principle worth writing down. Switching costs are high for enterprise software. That means customers already committed to either platform are unlikely to switch even if pricing changes, so the real competitive battle is exclusively at the acquisition stage. Different segments need different strategies depending on win rate and switching likelihood. Not all segments have the same value to defend. Some were marginal fits from the beginning and should be deprioritized, regardless of competitive dynamics.

Step five: rebuild. Concentrate sales resources on the top segments by historical win rate and feature usage — these are the segments where the product’s full value is used, and where the competitive comparison resolves in favor of the premium. For the contested price-sensitive segments, evaluate a deliberately simplified product tier at a competitive price point — not the full product, but a constrained version that competes on price while protecting the premium tier’s economics. This solution was invisible from within the existing framework, because that framework assumed all segments were equivalent and the competitive response had to be uniform. The decomposition revealed the segmentation logic the binary choice had been obscuring the whole time.

Building the Daily Practice

1. Maintain a running list of the beliefs you hold most confidently in your domain. These are the beliefs that feel like facts to you — the things you know for certain, the received wisdom you never question. Revisit this list quarterly. Pick one belief and trace it back to first principles. What evidence actually supports it? What would change it? What would it look like if it were wrong? This exercise is usually quick, and it often produces the unsettling discovery that some of your most confident beliefs are built on much shakier foundations than you realized.
2. Cultivate cross-domain reading as a deliberate discipline. The mental models that transfer most powerfully to you are the ones you encounter by reading outside your primary field. If you work in business, read evolutionary biology and physics. If you work in science, read military history and economic history. The encounter with a first principle from a distant domain — and the recognition that it maps onto a persistent problem in your own domain — is one of the most intellectually productive experiences available to you.
3. Maintain a questions file. Every time you encounter a question you can’t answer from first principles, write it down. Why does this market behave this way? Why does this organizational structure produce this specific result? Why does this type of person behave consistently this way across multiple contexts? Revisit the questions periodically. Work through them using the Feynman Technique. Your questions file is a map of the gaps in your own foundational understanding.

First principles thinking is not a tool you pull out only for major strategic inflection points in your life. Used only in those moments, it’s too unfamiliar to apply well right when the stakes are highest. The skill gets built through your daily practice on smaller problems, so that when the large problems arrive, the process is automatic for you and the reasoning is fluent.

Your practice has three components.

Closing those gaps is the work that makes you genuinely more capable, as opposed to merely more informed about the surface details of your field. Parrish’s observation from running Farnam Street for over a decade is the right place for you to land on this. The people who make the fastest progress in building genuine wisdom are not necessarily the most intelligent or the best educated. They’re the ones most honest with themselves about what they don’t know. That honesty is your prerequisite for first principles work. You can’t decompose a problem you’re not willing to examine. You can’t identify false assumptions you’re not willing to challenge. You can’t build on bedrock you haven’t been willing to find. The intellectual courage to say “I don’t actually know why this is true, and I need to find out” — that’s the entire first step of the Decomposition Protocol for you. Everything else follows from it.

The Social Dimension: First Principles Thinking in Groups

Everything we’ve covered so far treats first principles thinking as a solo intellectual practice for you. In reality, most of the decisions that matter to you — business strategy, organizational design, policy choices — happen in groups. And groups have social dynamics that systematically suppress first principles reasoning in favor of conformity, consensus, and the path of least social friction.

The social pressure against first principles challenges is strongest at exactly the moments when first principles reasoning is most needed. When a group has invested significant time and resources in a direction, the social cost of challenging its foundational assumptions gets highest right then. When a leader is visibly committed to a strategy, raising first principles objections to it risks you being perceived as disloyal or pessimistic. The social architecture of most groups you’ll sit in is optimized for implementation, not for honest evaluation of whether what’s being implemented is actually correct.

Irving Janis documented this in his 1972 study of groupthink — the phenomenon where the desire for harmony and conformity in a group actively overrides realistic evaluation of alternatives. His analysis of historical policy disasters — the Bay of Pigs invasion, Pearl Harbor, the escalation of the Vietnam War — showed a consistent pattern. Intelligent, experienced decision-makers suppressed doubts and failed to challenge foundational assumptions, because the social cost of dissent was too high inside strong group cohesion and strong leadership commitment. The intellectual failure wasn’t random. It was socially produced by environments that punished the exact behavior first principles thinking requires from you.

Here’s the practical response for you. If you’re in a leadership position, the single most powerful thing you can do to improve the quality of first principles reasoning in your group is to model intellectual humility publicly and visibly. Ask questions you don’t already know the answers to. Express uncertainty about your own positions out loud. Publicly update your views when presented with disconfirming evidence. Reward the team member who identifies the foundational flaw in the plan before it becomes the foundational flaw in the outcome. The social cost of first principles challenges drops to near zero when you, as the leader, demonstrate that such challenges are valued, not merely tolerated.

Jeff Bezos at Amazon institutionalized this through what he called the disagree and commit norm. It states explicitly that it’s not only acceptable but expected to disagree with a decision, voice that disagreement clearly, and then fully commit to the chosen path once the decision gets made. The sequence matters for you to copy: voice dissent before the decision, commit after. This structure gives first principles challenges a legitimate moment in your decision process while preventing the paralysis of endless post-decision second-guessing. It’s not perfect. The social pressure of hierarchical relationships still distorts what actually gets voiced. But it’s significantly better than the default norm of keeping disagreements private to avoid conflict.

First Principles and the Long Game

Munger described compound interest as interest earned on interest, compounding over time to produce outcomes that seem disproportionate to the original investment. It’s one of the first principles he applied to his own intellectual development as well as to his capital. The latticework of mental models compounds for you too. Each new first principle you genuinely understand increases the value of all the ones you already have, because it creates new intersection points for you — new ways to triangulate on problems using independent lines of reasoning from different domains.

This compounding makes your investment in first principles thinking genuinely long-term in its payoff structure. In your first year of serious practice, the gains are modest. You start questioning more of your own assumptions, you catch yourself reasoning by analogy in situations that call for something better, and you notice the gaps in your explanations more reliably. In years three to five, your latticework starts to be genuinely useful across domains. You start seeing the parallels between evolutionary selection pressure and competitive market dynamics, between the thermodynamic concept of entropy and organizational systems, between the psychology of loss aversion and the political economy of incumbent industries. In years ten to fifteen, your latticework is producing insights that are genuinely rare. Most people operating in your domain are still reasoning by analogy from the same set of inherited models. You’re triangulating from a library of first principles they don’t have access to.

Feynman described this compounding in his own intellectual biography. He spent the early years of his career building his foundations so thoroughly. He worked through every derivation, filled every gap, and never accepted a result he couldn’t himself derive. So when he encountered completely novel problems later, he had the tools to solve them from scratch. The investment in foundations paid off for him not in the early years when it felt slow, but in the later years, when the unusual problems arrived and everyone else was stuck.

The most important decisions in your life — the ones that will determine the trajectory of your career, your relationships, your financial position, your health — are not the routine decisions your analogy reasoning handles well. They’re the novel, high-stakes, ambiguous decisions that arrive without warning and demand sound judgment under uncertainty. First principles thinking is your preparation for those moments. It’s the intellectual infrastructure that makes good judgment possible for you when the pressure is highest and the stakes are real. Build it now, before you need it.

The Relationship Between First Principles Thinking and Your Creativity

There’s a widespread misunderstanding that first principles thinking is primarily analytical for you — a tool for breaking things down, not for building things up. This misunderstands the creative dimension of the method entirely. Feynman was one of the most creative physicists of the twentieth century. Munger built one of the most successful investment track records in history. Musk built multiple companies that conventional wisdom said were impossible. All of them used first principles as their primary creative tool, not just as an analytical one, and you can too.

The creative power of first principles thinking comes from what happens after the decomposition — step five of the protocol, the rebuild. When you’ve stripped away all your inherited assumptions and you’re working with only the bedrock truths, your solution space expands dramatically. The constraints that limited your options were never physical constraints. They were assumption constraints. The binary choice between two bad options was not a feature of reality for you. It was a feature of the framework you inherited. Remove the framework and your solution space opens up.

This is why the most creative breakthroughs across domains — in science, in business, in art — consistently come from people willing to question foundations everyone else took for granted. Darwin questioned the permanence of species, which everyone had accepted as a first principle. Einstein questioned the absoluteness of simultaneity, which Newtonian physics had treated as bedrock. Bezos questioned the assumption that physical retail required physical stores, and that service quality was inversely proportional to price. Every one of these was a decomposition that stripped away an inherited assumption and rebuilt from the actual underlying truths, which turned out to support a radically different structure than the conventional wisdom had produced.

The creative practice first principles thinking enables in you is not brainstorming. It’s structured decomposition followed by unconstrained reconstruction. Your deconstruction is analytical and rigorous. Your reconstruction is imaginative and unconstrained by anything except the actual first principles themselves. That combination — rigorous analysis of what’s true, plus unconstrained imagination about what those truths make possible — is the engine of genuine creative breakthroughs for you. It’s how you solve problems that everyone else has already given up on.

The problems worth solving in your life — the ones that define careers, build companies, and change fields — are almost always the ones that appear unsolvable from within the existing framework. That’s precisely why nobody’s solved them yet. They require stepping outside the framework. First principles thinking is your systematic process for doing that. The question isn’t whether the problems you face require it. The question is whether you’ll develop the discipline to apply it when they do.

Common Failure Modes: What Goes Wrong When You Try This

First principles thinking has failure modes worth naming explicitly for you. People who try the Decomposition Protocol without understanding them often give up when they hit one. They conclude the method doesn’t work, when the actual problem is a specific, correctable error in how they applied it. Here’s what to watch for in yourself.

1. Stopping too early. The most common error is treating an assumption as a first principle because you haven’t questioned it deeply enough to see that it’s an assumption. “People buy this type of product because it solves their problem” sounds like a first principle at first. Ask what problem specifically. Ask how severe it is, whether the current solution actually solves it, and whether the price reflects the actual severity of the problem in the customer’s hierarchy of needs. Push one level deeper on every “because” before you accept it as foundation.
2. Rebuilding with the same assumptions. After decomposing the problem, you’ll often rebuild using the same assumptions you identified in step two, just in a different arrangement. The structure changes but the foundations stay the same. This happens because challenging assumptions intellectually is much easier than actually operating from the world without those assumptions. If your rebuild looks essentially similar to what you had before, you probably didn’t actually remove the load-bearing assumptions. You just rearranged them.
3. Using first principles to rationalize a predetermined conclusion. This is confirmation bias applying itself to your own decomposition process. You start from the conclusion you want to reach, identify the first principles that support it, and call the result first principles thinking. It isn’t. Real first principles work produces conclusions you didn’t predict and sometimes don’t prefer. If your decomposition always leads back to what you already believed, you’re performing the ritual of the method without its substance.
4. Confusing the domain with the first principles. Every domain has its own jargon, its own canonical texts, its own accepted wisdom. Mastering these things is domain expertise for you, not first principles knowledge. The first principles of a domain are the foundational truths from physics, mathematics, biology, psychology, and economics that the domain is built on — truths prior to and independent of the domain’s own conventions.
5. Using first principles as an excuse to dismiss accumulated wisdom. This is the inverse of the previous failure mode. Because first principles thinking requires you to question inherited assumptions, you might use it as a blanket justification for dismissing expertise and accumulated experience. That’s not first principles thinking. It’s contrarianism dressed in intellectual clothing. Legitimate accumulated wisdom has been refined by feedback from reality over many iterations, and it’s often load-bearing even when you can’t immediately derive it from first principles. The correct posture is to question it, not dismiss it.

Shane Parrish captures the correct balance well for you. The goal isn’t to rebuild everything from scratch every single time. The goal is to know which parts of your thinking are actually grounded in bedrock and which parts are inherited assumptions you haven’t yet verified. The verified parts you use with confidence. The unverified assumptions you hold more lightly and update when evidence challenges you. The entire project is about knowing the difference, and that requires you to actually do the work of decomposition on the things that matter most to you.

The Compounding Return: Why Starting Now Matters to You

People ask some version of this question after almost every episode we do on a skill-building framework like this one: is it too late for me to start? Here’s the honest answer for you. The best time to start building a first principles reasoning practice was ten years ago, and the second best time is today. The compounding nature of the latticework means every year you wait is a year of compound growth you don’t get back. But it also means starting today puts you significantly ahead of where you’d be if you waited five more years.

Your practical starting point is simpler than you probably expect. You don’t need to read a hundred books before you can begin. You need to pick one important belief you hold confidently in a domain that matters to you — your career, your finances, your health, your most important relationships — and trace it back to bedrock this week. What’s it actually based on? What would change it? What would the world look like if it were wrong? That exercise, done honestly, produces a first-principles insight you can use immediately.

Then do it again next week. Pick another belief. Trace it back. Your practice is iterative and cumulative. After six months of weekly decomposition exercises on the beliefs that actually drive your most important decisions, you’ll have a qualitatively different relationship with your own thinking than you have today. Your assumptions will be more visible to you. Your derivations will be more traceable. Your bedrock will be more reliable. The decisions you build on it will be better.

Feynman, Munger, Parrish — none of them are arguing this is easy for you. They’re arguing it’s worth it, and that it gets easier with practice. Your first decomposition is the hardest, because you’re working against years of habit that treats inherited wisdom as bedrock. Each subsequent one gets slightly easier for you, because you’ve practiced the movement of stepping back from your assumptions and asking whether they’re actually true. Over years and decades, the movement becomes natural for you — not effortless, but natural, the way any genuine skill eventually becomes part of how you work rather than a special technique you apply on special occasions.

That’s the goal for you. Not to be someone who occasionally uses first principles thinking on the big decisions. To be someone for whom first principles thinking is the default mode — the natural first move whenever you face any problem worth solving. That person is more capable, more adaptable, more creative, and more reliably effective than the person reasoning by analogy from inherited frameworks. Building that capability is one of the most valuable investments you can make in your own development.

Here’s the falsifiability test worth carrying with you after this episode ends: what evidence would change this belief? If you can’t identify such evidence — if a claim feels true to you regardless of what the world actually shows you — it’s a belief, not a principle. A genuine first principle survives contact with disconfirming evidence, because it’s grounded in empirical reality rather than in how long you’ve held it. Watch especially for the trap of mistaking “this has always been true historically” for “this will always be true.” Historical consistency is not a first principle for you. The mechanism that produces the historical pattern is the actual principle, and it’s worth your time to go find it.

And if you’re wondering whether this only applies to business and technical problems, it doesn’t. It applies everywhere in your life, but it’s hardest in your personal decisions precisely because the inherited assumptions there are the most deeply embedded. Your career path, your relationship structure, your financial strategy, where you choose to live — all of these are heavily shaped by recipes your family, your culture, and your peer group wrote for you before you ever had a say. The decomposition process is identical for you here too: list the assumptions, challenge them, identify the bedrock truths, rebuild. The difference is that your bedrock truths in personal decisions are often psychological — what do you actually value, as opposed to what you’ve been told you should value? That question requires the exact same first-principles rigor you’d bring to any physics problem.

One caution before we close, because it’s real and you should know it going in: there’s a genuine risk of spending so much time on first principles analysis that you never actually act. Analysis paralysis driven by perfectionism in the decomposition phase is its own failure mode. The Decomposition Protocol is designed to produce better hypotheses for you, not certainty. The goal isn’t to think until you’re sure. It’s to think long enough to identify your load-bearing assumptions, challenge the critical ones, and build a foundation strong enough for you to act on. In practice, the six steps of the protocol take you one to three hours for a major strategic decision. That’s the investment. After three hours of genuine first principles work, you act. The alternative — acting without doing the work at all — is what produces years of compounding wrong assumptions, the kind that cost people like Tom and Rachel years of their working lives before they finally caught it.

The protocol is available to you now. The practice is yours to design. The only question left is whether you’ll start.

One last thing worth saying to you directly before we close. Nothing in this episode requires you to be smarter than you already are. First principles thinking is not a talent you either have or don’t. It’s a discipline, built the same way any discipline gets built — through repetition, on problems small enough that failing at it costs you almost nothing. You don’t need to decompose your entire career this week. You need to decompose one small, contained belief, and see what’s actually underneath it. That’s the whole starting move. Everything else in this episode is just what happens when you keep making that move, week after week, on problems that matter progressively more to you as your confidence in the process grows.

Think back to Musk for a second, because it’s easy to hear that story and conclude it only applies to men building rockets or founding companies you’ll never found. It doesn’t. The mechanism he used is available to you on Tuesday, on a problem as small as why your last three difficult conversations went badly, or why you keep abandoning the same habit at the same point every time. Trace it back. Find what’s actually true underneath the story you’ve been telling yourself about it. Rebuild from there. That’s the whole method, scaled down to the size of an ordinary week in your ordinary life, and it works exactly the same way at that scale as it did at his.

So pick the belief before you close this out. Not tomorrow, not after you’ve had time to think of a good one. Right now, while the shape of the method is still fresh in your head. Something small enough that being wrong about it costs you nothing except a little pride. Write down what you believe, write down why, and ask yourself the only question that actually matters: how do you know. Sit with whatever answer comes back. If the honest answer is “because that’s what I was told” or “because that’s how it’s always been,” you’ve just found your first crack in the foundation, and you didn’t need a rocket company or twenty years of research to find it. You needed five minutes and the willingness to ask yourself a question you’d normally skip.

We will be back next week.

You can explore more foundational thinking frameworks in the Mindset Tools library on this site, which maps the full architecture of the mental models this show covers. And for principles that operate under pressure and adversity specifically, the Resilience Principles section builds on many of the same foundations covered here.

Editorial StandardsCorrectionsMedical DisclaimerAbout Our ContentAffiliate DisclosureSite Map