First Principles Reasoning

Standing on a Floor You Think Is the Ground

Lesson 3.1 left you with a worksheet and a date on your calendar, and it ended with an admission I want to open with. Everything in that lesson assumed your premises were sound and only your reasoning was at risk. That is a generous assumption, and it is usually wrong.

Watch the failures closely and you find that the logic was mostly fine. The models were competent. The people were smart and often working hard. What failed was further down — a starting assumption that had been true once, or had never been examined at all, sitting under twenty years of excellent reasoning like a cracked footing under a well-built house. Nobody inspects the footing. Everybody inspects the walls, because the walls are what you can see.

Here is the tell. Someone questions a thing that has never been questioned, and the room does not produce an argument. It produces a feeling — a slight irritation, a sense that the question is naive or off-topic, followed by an answer that is not actually a reason. That is what a floor feels like when you mistake it for the ground. Bedrock does not need to be defended by tone. It just answers.

I have been on both sides of that. When I halved two employees' salaries with their consent, my logic was clean and my premise was that consent was the binding constraint. It was not. The binding constraint was a funding program's terms, which I had never gone down and read. Everything I built on that premise was reasonable and all of it was wrong, and the bill came out of my own pocket. Sound reasoning from an unexamined premise is not thinking. It is momentum with paperwork.

So this lesson is about descending. Not doubting everything, which is a posture rather than a method and which we will deal with honestly when we get to Descartes. Descending until you hit something that does not move, then building back up and seeing what survives.

Let me say what first principles reasoning is not, because the phrase has become fashionable and is frequently used to mean nothing more than 'I have decided to ignore precedent.' That is not it. Ignoring precedent is cheap, and mostly it is how people rebuild mistakes already solved and abandoned for good reasons. Reasoning from first principles is more demanding than ignoring the past, not less. It requires you to find out why the precedent exists, what problem it was built to solve, and whether that problem still exists — then to accept it on the merits or discard it on the merits. The person who cannot tell you why the rule exists has not transcended it. He just has not read it.

And here is the executive stake. Reasoning by analogy is fast and it is usually right, which is exactly why it fails catastrophically rather than gradually. It is right, right, right, right, and then the situation is not analogous and you have twenty years of momentum pointed at a wall.

The Four Floors

When you push on a belief and ask why, the answer will almost always be one of four things, and none is bedrock. Call them the Four Floors. Each feels like solid ground, each holds ordinary weight, and each gives way under load in its own characteristic way.

The first floor is Crowd: it is true because everyone I respect believes it. This is the most efficient floor and the most dangerous, because agreement is genuinely evidence — when a hundred competent people converge, that is a real signal. But it is evidence about people, not about the world. Ask how many of those hundred checked independently and you frequently find ninety-seven standing on the same three, one of whom has since changed his mind quietly. Municipal bond markets, mortgage markets, and denominations all run on this floor.

The second floor is Analogy: it is true because it worked in a situation that resembles this one. Analogy is the workhorse of practical reasoning and I am not asking you to give it up. I am asking you to make the resemblance explicit. Name the three features of the prior case that made it work, then check whether this case has all three. Usually it has two. The two are what makes the analogy feel right, and the missing third is what will hurt you. Growth playbooks imported from software into ministry, and policing strategies imported between cities, die on the missing third.

The third floor is Habit: it is true because we have always done it this way. This floor is easy to mock and hard to escape, and be fair to it before you tear it up. Habit is compressed experience. Chesterton's fence is the right instinct — if a fence stands across a field and you do not know why, do not tear it down until you find out. But the instruction is to find out. Most leaders use the fence as a reason never to ask, which inverts the point. The fence gets an explanation, not a permanent exemption.

The fourth floor is Authority: it is true because someone who outranks me said so. Not all authority is a floor — testimony from a competent, honest source is legitimate evidence, and no one can verify everything personally. The failure mode is precise: authority becomes a floor the moment you can no longer say what the authority's reason was. If you can restate the argument, you are standing on the argument. If all you have is the name, names do not bear weight when the situation changes.

So how do you know you have reached bedrock? By the shape of the answer. You have hit bedrock when the answer to 'why' stops being another 'because' and becomes one of four things: a law of nature that will not negotiate; a law of the land that binds whether you like it or not; arithmetic that has to balance; or a moral absolute. For this academy the last is settled by Scripture, and I will say that plainly rather than hedge — Scripture is the foundation, above tradition, above denomination, above the fathers, above church culture, and above whatever is currently fashionable. When Scripture speaks to a matter, that is the floor of the building. Everything else is filtered through it.

Everything above bedrock is a design choice somebody made, usually for a reason that made sense at the time. Design choices can be re-made. That is the point of descending: to find out how much of what you are defending is a law and how much is a decision that has gone unquestioned long enough to look like one.

Examine Everything Carefully

This is the academy's own method, and it did not come from a business book. It came from Scripture, which commands the checking rather than merely permitting it.

Paul writes to the Thessalonians: "But examine everything carefully; hold fast to that which is good" (1 Thessalonians 5:21, NASB). Read the halves together, because alone each becomes a vice. Examine everything without holding fast to anything and you get the perpetual skeptic, who has confused restlessness with rigor and never stands anywhere. Hold fast without examining and you get the man who cannot say why he believes what he believes and cannot survive the first serious challenge. The command is both. Test it, then commit to what passed.

And then Acts 17, the passage this curriculum is built on. Paul arrives in Berea and preaches. Luke records: "Now these were more noble-minded than those in Thessalonica, for they received the word with great eagerness, examining the Scriptures daily to see whether these things were so" (Acts 17:11, NASB). Sit with that. They are checking an apostle, running what he says against the text, daily. And Luke's verdict on the behavior is not that they were difficult, divisive, or insufficiently submitted to spiritual authority. He calls them more noble-minded than the people who did not.

That standard applies to you in both directions. It means you should check me. Nothing here is exempt — not the framework, not the case study, not the founder's conviction. And it means the people you lead should be able to check you without it costing them, because a leader who requires his conclusions to be received rather than examined has told you how much confidence he has in them. Notice also what the Bereans checked against. Not their preferences and not their culture, but a standard that outranked both them and Paul. Examination without a standard is opinion with more steps.

Philosophy formalized the same structure. In the Posterior Analytics, Aristotle argues that demonstration must proceed from premises that are true and primary, then shows why the chain cannot go on forever. If every premise required a demonstration, you would either regress infinitely or argue in a circle, and neither produces knowledge. So there must be first principles, not themselves demonstrated. That is not a weakness in the system; it is what makes the system possible. Every structure of reasoning you have built terminates in something you did not derive. The mature question is never whether you have undemonstrated starting points. You do. It is whether you know what yours are and can say them out loud.

Descartes pushed the method as far as it will go. In the Discourse on the Method and then the Meditations, he sets out to doubt everything that can possibly be doubted — the senses, the body, mathematics, the external world — to see whether anything survives, and finds one thing that cannot be doubted while being doubted: that he is thinking, and therefore is. It is the most famous descent in the history of philosophy, and as a demonstration that a person can go all the way down, it is unmatched.

Now let me steelman its limits, because a leader who adopts Cartesian doubt as an operating posture will destroy his organization. Descartes himself knew this. In Part Three of the Discourse, before beginning the demolition, he adopts what he calls a provisional morality — rules to live by while the rebuilding is underway, including obeying the laws and customs of his country and being resolute in action once he has decided. His reason is a builder's reason: before you tear down the house you live in, you must arrange somewhere else to live. Otto Neurath later put it better. We are like sailors who must rebuild their ship at sea, plank by plank, never able to put it in dry dock and start from the keel. You do not get to suspend operations while you audit your premises. You audit one plank at a time, standing on the rest.

Why Experience Both Helps and Blinds

There is a real body of research here, and I want to give it to you accurately, because it is routinely stretched into a claim it does not support — that experts are worse thinkers than beginners. They are not. If they were, none of us would hire them.

Start with chess, because that is where the work began. Adriaan de Groot found something counterintuitive: masters did not calculate far more moves than weaker players. What separated them was the quality of what came to mind first. Then William Chase and Herbert Simon made the mechanism visible. Show a position from a real game for a few seconds and ask the player to reconstruct it; masters do so dramatically better than novices. Now scramble the pieces into a random arrangement. The master's advantage largely collapses. He was never remembering twenty-four pieces. He was remembering a handful of meaningful configurations — chunks — and when the board contains no meaningful configurations, he has nothing to chunk.

That is what expertise is, and it explains both its power and its failure mode in one sentence: expertise is a library of compressed patterns that lets you skip most of the search. The skipping is the value. A veteran commissioner walks into a budget meeting, knows within ninety seconds which line is the problem, is usually right, and could not fully explain how he knows. He is chunking — the same phenomenon we met in Lesson 3.1 as recognition-primed decision making, described from the inside.

And now the failure mode, which has a name and a research history. Abraham Luchins ran the classic experiments in the 1940s using water-jar problems. Subjects were trained on a series all solvable by the same somewhat involved procedure, then given problems solvable by that procedure but also by a much simpler one. A large proportion kept using the long method, and some failed entirely on a later problem the trained method could not solve at all — one naive subjects handled easily. Luchins called it Einstellung: the mechanization of problem solving. The prior success installs a groove, and the groove is now the road.

Merim Bilalic, Peter McLeod and Fernand Gobet extended this to chess experts, in work published in Cognition, and the detail that matters is the eye-tracking. Presented with a position containing both a familiar winning motif and a shorter, better solution, strong players often reported the familiar one and stopped. Asked to keep looking, their eyes kept returning to the squares belonging to the familiar solution even while they sincerely believed they were searching elsewhere. Carry that finding, and carry its limits. It does not show that experts are generally blind. It shows that a good solution arriving early can suppress the search for a better one, and that the person doing it will not experience himself as having stopped searching. He will experience himself as having searched and found nothing.

Karl Duncker's older work on functional fixedness makes the point from another angle — people struggle to use an object in a role other than the one they have seen it fill, and the more familiar the ordinary use, the harder the reframe. Different literature, same shape. Prior structure is mostly an asset and occasionally a wall, invisible from the inside either way.

So the practical rule is not distrust your experience. Your experience is the most valuable asset you have. The rule is narrower: when a good answer arrives fast in a familiar-looking situation, that is precisely the moment to force one deliberate check. Ask what would have to be different here for my usual answer to be wrong, then look for that difference specifically rather than in general. The Bilalic result warns that looking in general does not work. Name the thing you are looking for, or your eyes will go where they always go.

The 'We've Always Done It This Way' Audit

Now watch an inherited premise fail under load, at the scale of a city, in living memory.

For roughly a decade before 1975, New York City covered a widening gap between what it spent and what it collected by borrowing short-term — tax and revenue anticipation notes, rolled over and reissued, with the proceeds increasingly funding operating expenses rather than capital projects. The accounting cooperated: expenses were pushed into the capital budget, uncollectible receivables were carried as revenue, and each year's shortfall was papered by next year's paper. Mayor Wagner's 1965 budget message defended borrowing for operating costs on the argument that the city's fiscal condition should not set the limits of its commitments to its people. This was not stupidity, and it was not simple cynicism. It rested on a premise that had been reliably true for a generation: the market will always refinance New York, because New York is New York.

That premise was a Crowd floor and an Analogy floor stacked on each other, and in spring 1975 it gave way. The banks stopped underwriting the city's notes. There was no dramatic event and no single villain — the buyers looked at the numbers and declined, and a mechanism that had absorbed every prior year's problem stopped absorbing. What followed came fast: the state created the Municipal Assistance Corporation in June, then an Emergency Financial Control Board in September that took effective control of the city's budget from its elected government. President Ford initially refused federal help, producing the most famous headline in the city's history, and signed seasonal financing legislation in December. Tens of thousands of city workers were laid off. The transit fare rose. In 1976 the City University of New York, free to its students since the nineteenth century, began charging tuition.

Here is the first-principles reading. Nobody in that decade was reasoning badly. They were reasoning competently from a premise — municipal credit is inexhaustible for a city of this importance — that nobody had checked, because checking it would have required saying out loud that the commitments exceeded the revenue. The bill was not created in 1975. It accrued for ten years and was presented in 1975. And the cost of correction fell hardest on people who had no part in the reasoning: the laid-off clerk, the rider, the first CUNY student in a century handed a bill.

That is what an unexamined premise costs at scale, and it generalizes to your organization exactly. So here is the audit, and it is a real exercise rather than a metaphor.

Take any standing policy, process, meeting, program, or budget line and ask four questions in order. What specific problem was this built to solve? Ask the person who was there; if nobody was there, that is itself a finding. Does that problem still exist in the same form? What would we build starting today, with no obligation to the existing thing? And finally, the question that separates an audit from an opinion: which parts of this are bedrock — legally required, physically necessary, arithmetically forced, or morally non-negotiable — and which are a design choice somebody made in a meeting you were not in? You will typically find the thing you assumed was a constraint is three-quarters convention around one genuine requirement. The requirement stays. The wrapping is yours to redesign.

The public-office version is the difference between policy inherited by precedent and policy justified from purpose. A candidate who can only tell you what the city currently does, or what his party currently holds, is standing on the Crowd floor and will be pushed off it the moment reality applies pressure. A candidate who can tell you what problem a program was built to solve, whether that problem still exists, what it costs, and what the strongest case against his own position is — that person can govern, and can change his mind in public without it looking like collapse, because everyone saw the reasoning the first time.

One last thing, and then Lesson 3.3. Everything I have described assumes that when you go down and check, you will see what is actually there. You will not, entirely. The mind has predictable ways of arriving at the conclusion it wanted while feeling like it examined the evidence — and the most dangerous ones do not announce themselves as bias. They speak in your own voice, using your own reasons, and they are most active precisely when the conclusion matters to you personally. That is where we go next.