First Principles Thinking

Why it matters

Almost everything you “know” about what a thing must cost, must look like, or must be is borrowed from how it has always been done — not from any law that says it has to be that way. Boil a problem down to the handful of facts you are certain are true, throw out everything you were merely told, and rebuild from there, and you often find that the wall everyone treats as solid was never load-bearing. The cost, the constraint, the “impossible” — most of it turns out to be convention wearing the costume of physics.

For example: for decades, every airline charged for a checked bag, served a tiny meal, and assigned you a seat, because that was simply what an airline was. A first-principles question — what does flying a person from A to B actually require? — separates the genuine fundamentals (a safe aircraft, fuel, a crew, a slot) from the inherited extras. Strip the extras and price each one separately, and you get the low-cost carrier: same physics, a fare a third of the old one, and a business the incumbents had assumed was forbidden by the nature of flying. Nothing physical had changed. The “rules” were just habits no one had re-derived.

What it reveals. Which of a problem’s constraints are genuine bedrock — physical law or proven fact, binding absolutely — and which are merely inherited convention, binding only because no one has questioned them.
How it changes the read. You stop asking “how do we do a better version of the usual approach?” and start asking “what is actually, fundamentally true here, and what would I build if I used only that?”
When to foreground it. A solution that feels clunky or expensive but goes unquestioned; a cost or constraint asserted as immovable without proof; a phrase like “that’s just how it works” where the explanation stops; a market or technology shift that may have voided old assumptions.
What you’d miss without it. That the design space is usually far larger than the conventional view admits — once the inherited assumptions are stripped, solutions appear that were never forbidden, just un-built.
Where it misleads. Genuine physical or economic laws can get waved away as “just convention” by an analyst who hasn’t understood them; and the framing can be invoked rhetorically — “I’m reasoning from first principles” — to dress up a pre-decided contrarian answer without any of the actual decomposition work.

How to invoke it in Ora

You want to understand how something actually works under the hood — how the parts produce the behavior you’re seeing — and to do that honestly you need to tell the real components apart from the assumptions everyone carries about what’s “obviously” there.

Name the phenomenon and ask:

“Explain the mechanism here — how do the parts and their interactions produce this behavior, at the principle level?”

This rides inside the Mechanism Understanding analysis, which locks a level of analysis, inventories the components with each one’s function, and accounts for how their interaction produces the whole’s behavior. The first-principles lens is one of the always-present points of view behind that work: it presses every item in the component inventory with the same question — is this a genuine fundamental of the system, or an inherited assumption about what’s there? — and it forces the boundary conditions to separate the constraints that are real physical law from the ones that are only convention.

One thing to know: phrases like how does this work, the mechanism, under the hood, what’s the principle, or explain the gears are what route you here. For step-by-step flow over time, a process map fits better; for the causes of a specific past outcome, a causal analysis — this is the structural how, the gears.

Give it the phenomenon and the behavior you want explained, along with whatever components you think are involved; the lens earns its keep by testing whether those really are the fundamentals, so don’t pre-filter them down to the “obvious” ones.

One thing Ora won’t do: accept “that’s just how it works” as the end of an explanation. It treats an unverified constraint as a question, not a fact — it asks what would actually make that constraint mandatory, and an analysis that strips assumptions without rebuilding a real account is flagged as unfinished.

How it works

There are two ways to figure out what something has to cost, and most of the time we use the lazy one without noticing. The lazy one is reasoning by analogy: we do it this way because that’s how it’s always been done, because that’s what everyone else charges, because last year’s number was X so this year’s is about X. It’s fast, it’s usually safe, and it is the default setting of the human mind. The other way is reasoning from first principles: you boil the thing down to the most fundamental truths you are sure are true — the physics, the proven facts, the things that genuinely can’t be otherwise — and then you build your answer back up using only those, refusing to import any of the inherited “everybody knows” along the way.

The cleanest modern illustration comes from the engineer Elon Musk, around 2013, trying to build an electric car. The expert wisdom on batteries was flat and unanimous: battery packs cost about six hundred dollars per kilowatt-hour, they always had, and they always would — so a cheap electric car was simply impossible. That’s an analogy: the future will cost what the past cost. Musk asked a different, almost childish question instead: what is a battery actually, physically, made of? The answer is cobalt, nickel, aluminium, carbon, some polymers, and a steel can. So he asked the next question: if you bought those raw materials on the metals market and nothing else, what would they cost? The answer came back around eighty dollars per kilowatt-hour. There it was. The gap between six hundred and eighty wasn’t a law of physics — physics permitted eighty. The gap was convention: current manufacturing methods, current supply chains, the accumulated way it had always been done. The expensive battery wasn’t forbidden. It was just un-built.

That is the whole method, and you can run it on almost anything. First, decompose. Take the problem apart and list every constraint and assumption baked into the usual approach. Second, sort. For each one, ask the decisive question: is this a genuine physical law or a proven fact — which binds absolutely, no matter how clever you are — or is it merely convention, which binds only because no one ever re-derived it? This sort is the hard part, and it’s where the discipline lives, because conventions are very good at disguising themselves as laws. Third, discard the conventions, keep the bedrock. Fourth, reconstruct — build a solution using only the bedrock truths as your building blocks, treating every conventional element as optional unless you can prove from scratch that you actually need it. The reconstruction is not optional; stripping assumptions and then walking away gives you a critique, not an answer.

Why bother, when analogy is so much faster? Because first-principles thinking is slower and harder than analogy every single time — and it is also the only route to a genuinely new solution when the conventional approaches are all trapped by the same hidden assumption. This is the quiet payoff that makes the effort worth it: the moment you strip away the inherited conventions, the space of possible solutions is suddenly far larger than the conventional view ever allowed, because most of what made the old space feel small was assumption, not law. The same problem, re-solved from the ground up, routinely produces an answer the conventional approach would have called impossible — not because the rules were broken, but because most of the “rules” were never rules.

The idea is ancient. Aristotle, writing his Physics around 350 BCE, gave it its name: first principles — the irreducible bedrock starting points from which all real knowledge of a thing is built, the points you cannot derive from anything more basic because there is nothing more basic. Two thousand years later, Descartes turned it into a deliberate procedure in his Discourse on the Method: doubt everything that can possibly be doubted, scrape away every belief that rests on mere authority or habit, get down to what cannot be doubted, and rebuild knowledge upward from that indubitable floor. Strip to bedrock; reconstruct from bedrock. The name for the whole move is, plainly, first-principles thinking.

Framework & implementation

This section uses Ora’s own terms for the parts of an analysis, so that if you open the actual mode and lens files they line up. Each is glossed in plain language on first use.

Pipeline execution

First-principles is one of the always-loaded mental models in the Mechanism Understanding analysis — it sits in the mode’s ANALYTICAL PERSPECTIVES block under “always loaded,” beside emergence, the map-territory discipline, Lakoff’s conceptual metaphor, Occam’s razor, falsifiability, feedback loops, and System 1 / System 2. It is not the mode’s method (Mechanism Understanding has no single required lens; its method is the locked-level component-and-interaction account); first-principles informs the read. The mode runs at Gear 4, Ora’s most thorough setting — a Depth analyst and a Breadth analyst work the mechanism in parallel, critique each other (cross-adversarial evaluation), and revise. (Note: the public Mechanism Understanding mode page is not yet built, so the “Foregrounded inside” host-link is absent for now; it heals automatically when that mode page lands.)

Honest host-fit note. The lens’s own file scopes it to problem-decomposition, assumption-audit, novel-solution-design, and constraint-verification. Mechanism Understanding is its public host, and the connection is apt: understanding how parts produce a whole requires decomposing to the genuine fundamentals — first-principles supplies the discipline of separating bedrock truths (the real components, the real constraints) from inherited assumptions, which the mechanism analysis needs in order to inventory components honestly rather than copying the conventional list of what’s “obviously” there. Its broader native use is wider than this host: breaking a problem down to its fundamentals in order to design genuinely novel solutions.

Where the lens engages. It activates on its Detection Signals — an existing solution that feels clunky but goes unquestioned; someone saying “that’s just how it works” with the explanation stopping there; a cost, complexity, or constraint asserted as immovable but never independently verified; a market or technology shift that may have invalidated old assumptions; the need to tell a real constraint from an assumed one. Its Application Steps are the decomposition protocol: state the problem clearly, list every assumption baked into the existing approach, classify each as physical law, proven fact, or mere convention, discard the conventions and keep only bedrock truths, and reconstruct a solution using only those bedrock truths as building blocks.

What it contributes to the analysis. It disciplines two of the mode’s sections in particular. It interrogates the Component inventory with function — pressing whether each listed component is a genuine fundamental of the system or an inherited assumption about what’s present, which is exactly the support the mode’s CQ2 needs against the component-inventory-without-function failure (a component copied from convention rather than verified rarely earns a real functional role). And it sharpens the Boundary conditions, forcing the analyst to mark which constraints are physical law (binding absolutely) versus convention (binding only by habit) — the bedrock-versus-assumption sort applied to the mechanism’s limits, which backs the mode’s CQ4 against scope-overreach.

Cross-adversarial evaluation. At Gear 4 each analyst’s reading is critiqued by the other, which catches the lens’s signature failures, keyed to its Critical Questions and Common Failure Modes: Rhetorical first-principles (the framing is invoked but no actual decomposition is shown); Physics-as-convention (genuine physical or economic laws dismissed as mere convention); and Reconstruction ignored (assumptions stripped but no replacement solution built — the analysis stops at deconstruction). The evaluator presses the core check: was the decomposition actually performed — assumption list and bedrock-truth set on the table — or is “first principles” being used as a label for a pre-formed conclusion?

What the analysis will not do. It will not accept an unverified constraint as bedrock; will not let a genuine physical or economic law be discarded as convention without cross-checking; and will not treat a stripped-down critique as a finished analysis — the reconstruction from bedrock truths is required, not optional.

Origin and evidence

The idea runs from ancient philosophy into modern engineering. Its origin is Aristotle’s Physics, Book I (c. 350 BCE), which introduced the archai — the first principles, the irreducible starting points of knowledge — and argued that in any systematic inquiry, real understanding comes from reaching the elements and causes that cannot themselves be derived from anything more basic. Its decisive methodological statement is Descartes’ Discourse on the Method (1637), where the program of doubt becomes explicit: reject every belief that can be doubted, reduce to what is indubitable, and reconstruct knowledge from that foundation — methodological reduction to bedrock, then rebuilding. The contemporary engineering application is associated with Elon Musk (around 2013), who described decomposing problems to their physical fundamentals — most famously the battery-pack cost broken down to raw-material prices, and the same move applied to rocket costs — to show that supposedly fixed costs were conventional, not physical. A complementary technique worth naming is Charlie Munger’s inversion: rather than asking how to succeed, ask what would guarantee failure and avoid that — a different angle on the same anti-convention discipline of refusing to take the standard framing for granted.

Applications and common uses

First-principles thinking is a working tool wherever an inherited approach is treated as fixed but its foundations have never been checked.

Engineering and product design. Its modern native ground: decomposing a product to its physical fundamentals and material costs to find that the “impossible” cheaper or better design was only un-built, not forbidden.
Cost and pricing analysis. Asking what a thing fundamentally requires to be made or delivered, and pricing up from there, rather than anchoring on what it has historically cost or what competitors charge.
Strategy and business-model design. Stripping an industry’s “this is just how our business works” conventions to bedrock customer needs and real constraints, then rebuilding the model — the move behind many category-redefining entrants.
Constraint verification. Auditing a constraint asserted as immovable — a budget, a regulation read as physics, a technical “limit” — to establish whether it’s genuine law or unexamined habit before accepting it.
Scientific and analytical reasoning. Refusing “everyone knows” as evidence, reducing a question to what is actually established, and reconstructing the inference from there.

In every case the payoff is the same: the genuine constraints are separated from the inherited ones, the conventional constraints are discarded, and the solution is rebuilt from bedrock — opening a design space the conventional framing kept hidden.

Failure modes and when not to use it

The lens’s characteristic ways of going wrong are catalogued in its Common Failure Modes:

Rhetorical first-principles. Invoking the framing to justify a pre-decided contrarian answer without doing the decomposition. The tell: the phrase “from first principles” appears but no assumption list and no bedrock-truth set ever do. The correction is to require both as outputs before the analysis is accepted.
Physics-as-convention. Dismissing a genuine physical or economic law as “just convention” because the analyst hasn’t understood it. The tell: a constraint that is actually mandatory gets discarded, and the reconstructed solution quietly violates it. The correction is to cross-check the bedrock-truth set against domain expertise before discarding any constraint.
Reconstruction ignored. Stripping the assumptions but never building the replacement — stopping at deconstruction. The tell: a satisfying teardown of the conventional approach with no actual solution standing where it was. The correction is to complete the reconstruction step before delivering.

When not to reach for it. When the conventional approach is already near-optimal and its constraints really are physical, first-principles decomposition is expensive overhead that returns the answer you started with. When the decision-maker lacks the time or authority to act on a substantially different solution, the analysis is academic — it will surface a better design no one can adopt. And the lens tells you which constraints are real and opens the larger design space; it does not by itself hand you the novel solution — the reconstruction is genuine creative work that the decomposition enables but does not perform.

Mechanism Understanding — the analysis this lens informs; explains how parts produce a whole’s behavior, and relies on first-principles to test whether its component inventory holds genuine fundamentals or inherited assumptions.
Emergence — the sibling always-loaded model in the same analysis: where first-principles strips a problem down to bedrock truths, emergence builds the whole up from the local rules of the parts.
Falsifiability — the kindred epistemic discipline: first-principles asks what would make a constraint genuinely mandatory, just as falsifiability asks what would make a claim wrong — both refuse to accept an assertion on authority.
Five Whys — the procedural relative: where Five Whys drills down to a problem’s root causes, first-principles drills down to its root assumptions, stripping inherited convention to bedrock.