The Philosophy of Everything 2e — Part I, Ch 3: The Knowing Process

Feeling sure proves nothing. A belief you were told a hundred times feels exactly like one you tested, and most people only learn which was which when the floor gives way.

In brief: In this chapter I follow knowledge from its rawest form, a belief nobody has tested, up to the theories that whole generations have leaned on without them giving way. Then I look at the roads by which we come to know anything at all, and at the quiet ways each road fails. It matters because most of us live on beliefs we have never tested, and the weak ones fail us slowly, over a whole life, without our ever seeing why.

Once I had taken apart the machine that does the seeing for me, I knew something. The senses and the brain were doing work I had not been bothering to look at. So the next thing I had to figure out was which of the things coming out of the machine I could lean on. Because that is what knowledge is, in the end. Knowledge is whatever holds your weight when you put weight on it. A belief is something you have not yet leaned on. A hypothesis is a belief you are about to lean on, on purpose, to see what happens. Knowledge is what is still standing after you have leaned. And a person who lives without ever asking which of his beliefs can hold his weight is a person who falls a lot. He falls slowly, over the course of a life, without ever quite seeing the floor coming up.

I used to think knowing was a feeling. If I felt sure of a thing, I called it knowledge. If I felt unsure, I called it doubt. The level of conviction in my chest, how sure I felt, was the only check I bothered with. And conviction, it turns out, is one of the easiest things in the world to feel for the wrong reasons. The certainty you carry about something you have actually tested feels the same as the certainty you carry about something you have only been told often enough. You cannot tell them apart from the inside. That is why knowing has to be done with something more than your feelings. It has to be done with checks that another person could run themselves and arrive at the same answer. Without that, you have a feeling. Feelings are not nothing. Many of them turn out to be right. But they are not knowledge until they have walked through a check and come out the other side.

So let me walk you through what I came to see as the chain. It runs from the rawest material in your head to the most polished. At the raw end is something you have decided is true without ever testing it. At the polished end is something many generations of people have leaned on without it giving way.

At the bottom of the chain is a belief. A belief is something a person has decided is true without ever testing it. Some beliefs turn out to be right when you do test them. Some turn out to be wrong. Beliefs are not enemies. They are starting points. A child believes the sun moves across the sky. So did the smartest people in every village for thousands of years. The belief was a reasonable starting point. The trouble is not in having beliefs. The trouble is in mistaking them for knowledge. It is in leaning on them as if they would hold, when they have never been tested for weight.

A hypothesis is a belief stated clearly enough that someone could go out and check it. Someone says, "I think gravity bends light." That is a hypothesis. The claim is shaped in such a way that you can imagine going out and finding the answer. If the claim cannot be checked even in principle (that is, even with perfect tools and all the time in the world), it is a belief, not a hypothesis. Most religious claims sit on this line. They sound like hypotheses. They are not. A claim that can never be disproved, no matter what happens, is not really a claim about the world. It is a feeling dressed up in a sentence. The technical word for what separates a real hypothesis from a feeling is falsifiability. A hypothesis has to allow, in principle, for some piece of evidence that, if it came in, would force you to give the hypothesis up. Without that, there is nothing to test.

A valid hypothesis is a hypothesis that has been tested and survived. The light-bending prediction was tested early in the last century (in 1919, during an eclipse of the sun), and the light did bend, and the hypothesis was promoted. It moved up a step. It became something more solid than a guess.

A theory is what you get when several valid hypotheses, each tested in its own right, lock together into one coherent description. The theory of relativity is a knot of many tested hypotheses, all checking out, all pointing in the same direction. A theory is the working unit of human knowledge. It is the strong end of the chain, the part that holds weight. Newspaper writers sometimes use "theory" the way ordinary people use "guess": "that's just a theory." This is wrong, and it has helped a lot of bad ideas survive longer than they should have. A theory in science is the strongest end of the chain, not the weakest.

Knowledge is the body of theories that have been built up this way. It belongs to no one. It is the shared deposit of generations of people doing the slow checking, mostly people you have never heard of, mostly without recognition or reward. It is what the species can stand on. It is the closest thing our species has to a treasure.

And a fact, finally, is a short usable piece of knowledge. It is a smaller, simpler statement that an ordinary person can pick up and put to work without having to follow the whole chain that produced it. The full theory might be too big to fit in a head. The fact is the postcard version. "Water boils at one hundred degrees Celsius at sea level" is a fact. The whole edifice of thermodynamics, the great building of the science of heat, sits behind that one sentence. But you can use the sentence without knowing the edifice.

I want to say one thing, sharply, about how knowledge is sometimes defined in philosophy textbooks. The textbook definition is "justified belief." To be fair, most textbooks actually say "justified true belief", and the word "true" closes part of the hole I am about to describe. Philosophers have argued over whether even that is enough since Edmund Gettier's short paper of 1963. I do not like this definition. It is not strong enough. It leaves a hole. Anyone can claim that their belief is justified, and the hole lets every kind of pseudo-knowledge walk in, things that only look like knowledge.

The flat-earth societies are full of people who believe they have justified their belief. They have not. They have rationalised it. There is a difference between rationalising and checking. Rationalising is putting reasons around a belief you have already decided to keep. Checking is putting the belief in front of a method that could destroy it, and seeing whether it survives. Knowledge is what survives the second thing, not the first. Without the survival, you have a hunch with paperwork.

Two small checks have to ride along with anything we call knowledge. The first is that the underlying facts have to be clean: free of confusion, free of dressing-up. The sun produces sunlight. The day is the time when there is sunlight where you are standing. Those are clean. The second is that the link between any two facts has to be honest. The day has sunlight because the sun is shining on this side of the earth. The link is accurate. Plenty of human reasoning falls apart not because the facts are wrong but because the links between them are wrong. People put together two true things and conclude a third that does not follow.

Once I started looking at the chain this way, I noticed something else. Every act of knowing, at any link in the chain, has the same four pieces underneath it. There is the person doing the looking. There is the thing being looked at. There is the conclusion the person comes to. And there is the way the person checked their conclusion. The observer, the idea, the result, and the means of checking. If any one of the four is missing, what you have is not knowing. It is something that looks like knowing.

I want you to keep these four pieces in mind, and here is why. Once you can see them in any claim somebody is making, you can see why so much of what passes for knowledge in ordinary life is missing one piece. Usually it is the last one. The conclusion is there. The observer is there. The thing being looked at is there. The way it was checked is gone, or was never there to begin with. Nobody mentions this. The conclusion just gets stated, with a confident face, and the listener is expected to accept it. Most of what is wrong in public conversation, I have come to think, is the missing fourth piece, walked around as if it were nothing.

Here is an example small enough to live in a sentence. A brother is walking down the road. He sees his sister lying on the ground next to her bicycle. He sees her through his own eyes. The thing being seen is his sister, on the ground. The conclusion he draws is that she has fallen. The way he checked was direct sight. That is the structure. Then the sister looks up and tells him she was only resting. The check fails. The conclusion he had, that she had fallen, turns out to be wrong. The other three pieces were fine. The means of checking gave him the wrong answer this time. This happens. Direct sight is not as trustworthy as it feels. That is what I want to slow down on next.

The most ordinary kind of knowing is the one you do every day without thinking about it. You look. You smell. You taste. You touch. You hear. Whatever your senses bring you, you take it in and call it knowing. Most of the moment-to-moment business of being alive runs on this kind.

Yousuf comes home from work. He recognises his own house because the shape of the building matches the shape in his head from a thousand previous evenings. His daughter Sana runs to him and presses her face into his neck, and he knows she loves him because he can feel her arms holding on. He drinks a hot drink and knows it is coffee, not tea, because the bitterness matches the coffee-print in his memory. He smells gas in the kitchen and knows the knob has been left on by Sana, who is still learning. He goes and shuts it before something terrible can happen. He hears his boss on the phone, and the tone is colder than usual. He knows from years of listening to that voice exactly what kind of trouble he is in. None of this needed a long argument. The senses brought him the information. He took it.

This is what most of life is, most days. Direct knowing, running quietly underneath, telling you what is happening right now in the small space around your body. Useful. Fast. Familiar. And not, as it turns out, something you can rest the whole edifice on.

Because direct knowing fails. It fails more often than people are willing to admit. It fails for three reasons that are worth slowing down on, because each of them shows up every day. They show up in every person's life, including the people who think themselves the most careful.

The first reason direct knowing fails is that the perception itself can be wrong. Yousuf goes for an evening walk. He sees a long, dark shape coiled on the path and steps back, fast, because his brain has filed the shape under "snake." A second look reveals it is a length of rope. The brain ran ahead of the slower checking machinery and gave him an answer, because the cost of being wrong about a snake is high. We do this all the time. The mind hurries. Sometimes it hurries in the wrong direction. Most disagreements between people, when you slow them down, turn out to be two minds that hurried to different snakes from the same rope.

The second reason direct knowing fails is that the senses themselves are not always working properly. A person with jaundice will see a white flower as yellow. That is the classic example, though doctors would tell you it is rare. Jaundice turns the whites of the eyes yellow, but only in a few severe cases does it turn the whole world yellow too. The flower has not changed. The eye has. A person with a blocked nose will not smell a gas leak. The leak is real. The smell is just not arriving. Older people often see worse than they remember seeing. Their hearing softens without their noticing. The brain makes up for it, for as long as it can. Then one day a road accident makes the compensation visible, and you can see how much the brain had been covering for. The instrument was not what it used to be, and the person had not bothered to find out.

The third reason direct knowing fails is more dangerous than the other two, and it deserves more time. Sometimes the mind sees things that are not really there at all. The full set of senses can deliver an experience that has no actual cause out in the world. I am not talking only about hallucinations. I am talking about the more common thing. It is that the mind, fed enough of a particular story over enough time, begins to act as if the story is the world.

Here is one of the most chilling examples I know of, and the original book (the first edition of this one) reaches for it too. There is a story (a novel, I think) about a famous actress called Sharon Fields. She has a coquettish public image, playful and flirty. Her publicity team, eager to keep her name in the news, has taken that image and run with it. They have put out stories that make her sound not just flirty but available, eager, waiting for any admirer who turns up. The stories are inventions. She has never said she wants what they say she wants. But four men, including a supermarket manager called Adam Malone, take the stories for truth. They become convinced that the actress wants what the publicity says she wants. So they kidnap her. They rape her. Her real protests, when they finally hear them, fall on ears that have already decided what the truth is. The publicity has done its work.

Adam Malone is a criminal. He is also, in a strange way, a casualty (a victim) of a lie marketed as the truth. The mind of a person who has been fed a fiction long enough begins to lose the ability to tell the fiction from the world. This is not just a story in a book. It is what happens to ordinary minds under the steady pressure of repeated false stories. Our minds will believe what we hear most often. The world is full of people, and groups of people, who know this and use it.

I want to be honest about something here, because I am asking you to take this seriously. I have caught myself, more than once, holding a view about a person or a group of people. On a careful look, the view turned out to come not from any direct experience of mine but from things I had read or heard repeated. The view felt like knowledge. It was not. It was a fiction I had absorbed without noticing, then defended as if it were my own conclusion. The work of catching yourself doing this is uncomfortable. There is no other word for it. But the alternative is being a person whose mind belongs to whoever was loudest near them when they were young. That is not a life I want, and it is probably not the life you want either.

I want to be careful, though, not to make any of this mean that imagination is the enemy. The same faculty, the same power of the mind, can build a false picture of a film star out of a publicity team's lies. It can also build a useful picture of a possible solution to a problem you are stuck on. Imagination is how scientists guess what to look for next. It is how an ordinary person figures out a route home through a city they have never been to before. It is how a child invents a game. Used carefully, with the discipline of going back and checking against the world, imagination is one of the best tools we have. We should use it more, not less. The fault in the Adam Malone story is not the imagining. The fault is the not-checking. A mind that imagines and then checks is the mind of an inventor. A mind that imagines and never checks is a mind on its way to a kidnapping.

There is a fourth thing to say about direct knowing, and it is small but worth a paragraph. Sometimes there is no perception at all. A boy finds a bucket of clear, odourless liquid in the storeroom and thinks it is water. He uses it to wash the cricket pads. The pads start to fall apart. The liquid turns out to be dilute sulphuric acid. The boy was not wrong. He was ignorant. He did not have the means to tell water from acid. The two look the same, smell the same, weigh roughly the same. Ignorance is not a mistake. It is an absence of information. The way out of ignorance is not better instincts. It is more information. Most of education is, at the bottom, the slow filling-in of the places where you cannot yet tell water from acid.

One more thing about direct knowing, because it ties to something I want you to remember from the last chapter. Some of what your senses give you is what we called subjective: it lives inside one head, and it cannot be settled by going outside that head. A film that moves you to tears may leave the person sitting next to you bored. The film has not changed. The two of you have. Each mind brings its own history to the looking, and the looking arrives shaped by that history.

The only way to push towards something more public, more objective, with this kind of perception is to ask many people what they felt and look at the average. A doctor decides what blood pressure counts as normal by taking the readings of a million people and finding the middle of the curve. I am simplifying here. Today doctors mostly draw the line for high blood pressure where large studies show heart attacks and strokes becoming more likely, not at the middle of the curve. We can do something similar with subjective experiences if we have to. We can find out, with care, what most people feel and call the common ground objective. It is not a perfect method. It is the best one we have for inner perception.

Together, the three failures and the ignorance and the subjectivity tell me one thing: direct perception, useful as it is, cannot be the floor you stand the whole house on. The floor has to be wider. We need other ways in.

Direct knowing only reaches what is close enough to touch. For everything else, I had to find other ways in. For the wind, for gravity, for the past. For what is happening in another country, and for what is happening in another person's head. I had to learn to know things sideways.

There are three big sideways roads. You can infer, that is, work backwards from clues to causes. You can ask other people, and you can argue with them. You can run a careful test and watch what happens. Three roads, three sets of tools. Let me walk you through each one.

When you smell smoke, you do not wait for someone to tell you there is a fire. You walk yourself, in a single small move of the mind, from the smoke to the fire. This is inference. It is the most everyday tool of the sideways set. We do it hundreds of times a day, mostly without noticing.

Imagine a group of friends gone hiking who lose their way. They spend the night in a makeshift shelter, something rough they have put together themselves. In the morning, they see a thin column of smoke rising over the trees in one direction. From the smoke they conclude their other friends, who set out the day before, have made camp over that ridge.

Look at how the inference holds together. There is the thing they actually see: the smoke. There is the reason they are drawing the conclusion they are drawing: no one else is out here at this time of year. There is the example, the general principle from a life of observations: smoke comes from fire, and fire here means people. There is the application: they should walk towards it. And there is the conclusion they reach by stringing the first four together: those are our friends. Fact, reason, example, application, conclusion. Five steps. They run together so smoothly in the moment that you cannot see them. But each step is doing work, and any one of them can be wrong, and when one of them is wrong the conclusion goes with it.

The step most likely to break is the third one: the example, the assumed connection between two things. There are three kinds of connection that inferences ride on. It is worth knowing all three, because they show up in different problems.

The first is the connection between a whole and its parts. Hand someone a stainless-steel water bottle. They turn it in their hand. They know what it is made of. They learned in school that stainless steel is iron, carbon, nickel, and chromium. Knowing the whole, they have inferred the parts. The reverse also works. If a chemistry teacher tells you that two atoms of hydrogen have been joined with one atom of oxygen, you can infer that what they have made is water. Knowing the parts, you can infer the whole.

This little two-way move underlies two great tendencies in human thinking. There are people who try to understand something by breaking it down into its smallest pieces. Call this the reductive way. There are people who try to understand it by looking at all the pieces together and watching how they cooperate. Call this the holistic way. Both are useful. Neither is enough on its own. The reductive way has built much of modern medicine: name the part that is broken, treat the part. The holistic way is older. It is at the heart of the traditional systems of healing in many parts of the world, including Āyurveda, the traditional medicine of India.

The reductive way works wonderfully for a broken bone. It works less well for a depressed person. The cells in your body are not the parts of a car. They speak to each other. They share information. A treatment that ignores the conversation between them ends up fixing one cell and breaking another. A good doctor, in my view, uses both routes. Such a doctor knows what each part is doing, and what the whole arrangement is doing, and treats both. A medicine that is only reductive is a medicine that has stopped listening.

I want to say a hard word about animal experiments in this connection. A rat in a cage does not know it is in a cage. It cannot choose to exercise because it has noticed it is getting heavier. The mind of a rat is not running the same program as the mind of a person. We can change our bodies by changing our minds. A rat cannot. So studying us by studying it has a ceiling. There is a place for animal testing. There is also a point at which the testing tells us less than we keep pretending it does. At that point the suffering of the animals is greater than the value of the information we get. I think we passed that point some time ago and have not been honest about it.

The second kind of connection is between a substance and its qualities. You are handed a glass of water. You can tell in a moment that the substance has properties: clear, tasteless, odourless, wet, neither very heavy nor very light. From the qualities you can infer the substance. From the substance you can infer the qualities. Either direction works. You see a key in someone's hand and you can infer it opens a door, even if you do not know which door. You see a door that needs opening and you can infer that what you need is a key, even if you do not know which key. This is the workhorse of practical thinking, the tool that does most of the heavy work. Most household problems get solved this way.

The third kind of connection is between an individual and the class it belongs to. Human beings have been classifying living things for thousands of years. The system we use now puts every living thing into a hierarchy of groupings: kingdom, phylum, class, order, family, genus, species. A cyclist on a forest road sees a large striped cat, and the chain runs in a moment. This is a tiger. Tigers are part of the genus Panthera. Panthera are part of the family of cats. Cats are carnivores. Carnivores eat meat. This animal is therefore not somebody I want to share the path with. The whole chain rides on classification. Each level lets the cyclist infer something about the level above. None of it was directly verified in the moment. It was inferred from the class the individual belongs to.

These three connections (whole and part, substance and quality, individual and class) are the rails that inferential thinking runs on. When inference goes wrong, it is usually because one of these connections has been wrongly assumed. A whole has been broken into wrong parts. A quality has been assigned to the wrong substance. An individual has been put in the wrong class. Most prejudice, to take one example, is the third kind of error scaled up. It is an individual being assigned the imagined qualities of a class they have been wrongly placed in. The reasoning machinery is fine. The premise is rotten. (A premise is the starting point that the reasoning stands on.) Rotten premises do not announce themselves. They have to be looked for.

The next route I had to think about was the one that runs through other people. There are two kinds of moves along it. You can take what somebody else tells you: that is testimony. You can argue with somebody about a claim: that is debate. Both are old. Both are essential. Both can be poisoned.

When the lost hikers finally reach the camp, the guide tells them a pack of wolves has been seen circling the area at night. They should not leave the camp after dark, he says. The hikers have not seen the wolves. They take the guide's word for it. They stay in. The next morning they are alive.

This is testimony. One person tells another what they know. The other person believes it without having to see it themselves. Testimony is the largest mechanism of knowledge transfer the human species has, the main way knowledge passes from one person to the next. It is older than writing, older than school, older than the village. Our ancestors survived by telling each other what they had learned. And our species ended up at the top of the food chain because our testimony was better than the testimony of any other animal. Every parent who tells a child not to cross the road without looking is using it. Every teacher in a classroom is using it. Every old person passing on what they learned from older people before them is using it. It is one of the most powerful things human beings do.

It is also one of the easiest things to break. For testimony to actually work as a route to knowing, two things have to hold. The first is that the person telling has to have actually got the information right. The second is that they have to tell it clearly enough that the listener gets exactly what was meant. If the guide says, "There may be some wolves around, I'm not sure, you might want to be careful," the lost hikers may decide to risk an evening walk. The vagueness kills the testimony. Good testimony is information that is right, said clearly, with no confusion left lying around for the listener to fall over.

But testimony has a dark side, and the dark side is most of human history. Most of the worst things people believe were given to them by other people they trusted. Think of the people who speak for God in your house: your priest, your imam, your guru, your grandmother who tells the old stories. They are passing on testimony every time they speak. The testimony has often been handed down through many generations of people who themselves never checked. By the time it reaches you, the original observation, if there ever was one, has been edited and re-edited beyond recognition. The testimony is loud. It comes wrapped in love. It comes wrapped in authority. It comes wrapped, sometimes, in threats: that if you do not believe what you are told, terrible things will happen. None of this changes what testimony is. It is information from somebody else, which still has to be checked before you can call it knowledge.

Not all bitter testimony is poison, though. Some of it, the testimony you most need to hear, comes in a form your mouth wants to spit out. There is an old image for this kind of advice. It is called gooseberry juice. Bitter on the tongue, good for the body. The friend who tells you the partner is wrong for you. The teacher who tells you the dissertation is going nowhere. The aunt who tells you the way you are raising the child is hurting them. We resist, every time. Years later, sometimes, we look back and realise the bitter advice was the right one, and we wish we had listened sooner. Some of the most important testimony you will receive in your life is the kind you would have paid to avoid hearing.

The internal experiences of meditation are the other place where testimony is the strongest evidence you have. These are not things a scientist can verify by hooking electrodes to a rat. The rat is not having the experience. So when someone whose life looks calmer than yours tells you the practice has done something for them, you take their word as a starting point. You try it. You watch what your own life does. If it benefits you, you keep going. If it does not, you stop, or try a different practice. The point is that the test still ends with you. The testimony was the door. The walking is yours.

The other half of the route through other people is debate. Two people, holding different positions on the same question, arguing it out, with the genuine intention of getting closer to the truth.

A good debate is one of the most beautiful things human beings can do together. One side proposes a position. The other side argues against it. Each side has to listen. Each side has to be willing to change its mind when the other is right. Each side has to be willing to admit when its own argument has run out of room. At the end of a real debate, both sides have moved. Either one wins, or they meet somewhere in the middle, or they agree the question is not yet ready to be decided. Any of those three is a victory. The mind has been improved on at least one side. Sometimes on both.

There is also a fake version of debate. Most of what passes for debate on television is the fake version. The participants do not want to learn anything. They want to win. They refuse to acknowledge a point even when the point is plainly true, because acknowledging it would cost them a face they have decided to protect. They lie, sometimes outright, when they think they can get away with it. They appeal to emotion when their reasoning fails. They change the subject when cornered. The whole performance is a kind of theatre. The watching audience comes away with no new knowledge but a strong feeling about which side seemed more confident. Confidence is not truth. It is the loudest thing in the room, often, and the easiest to mistake for it.

This is one of the things that makes democracy harder than it has to be. Most politicians, I have come to think, do not believe arguments win elections. They believe emotions do. And the sad thing is they are not wrong. Most voters do not bother to do the testing. The voter feels good or bad about a candidate, and the feeling decides the vote. The candidate's actual claims sit untested.

There is no quick solution to this. The slow one is to strengthen the people whose job it is (or should be) to actually do the testing. Someone has to be the torchbearer of truth in a democracy, and it cannot be the politicians. It has to be honest journalists. Scientists willing to talk about their work in public. Intellectuals who have made the slow, unglamorous work of checking into their daily practice. When this group is strong, the public has somewhere to look when politicians lie. When this group is weak, the politicians win every argument by default. A democracy is only as good as the people in it doing the checking. There is no shortcut around that.

Now, within an honest debate there are still two ways to win. You can show, directly, that your side is right. Or you can show, indirectly, that the other side leads to absurdity, to something that plainly cannot be.

The direct way is more satisfying. To show that the earth is round, you can point to the way an object stays the same size as you move around it. The sun looks the same size whether you see it at noon or at sunrise. A flat surface would not do that. You can also point at the photographs from space, the way ships disappear bottom-first over the horizon, the way the moon's shadow falls during an eclipse. (In an eclipse of the moon, the shadow that the earth throws on it is always round.) Several lines of direct evidence, each one independently strong, locking together.

The indirect way works by showing that the other side's position, if taken seriously, leads somewhere absurd. If the earth were flat, every part of it would be in day at the same time, or in night at the same time. But the sun is rising over Tokyo while it is setting over London. The flat-earth picture cannot account for this. The fact that it cannot is a kind of proof that the picture is wrong. You do not need to prove your own side directly. You only need to prove the other side leads to impossibilities. The conclusion follows.

A word here about doubt, because doubt is the engine of all of this and most people are taught to fear it. Sometimes you look at a question and you cannot tell which way to go. The evidence is mixed. The argument has not yet decided itself. You hold two possibilities at once and the mind does not settle on either. This is incomplete knowing. It is not a failure. It is the mind being honest about what it does not yet have. Most beginners in any field are made uncomfortable by this state. They have been taught that not knowing is shameful. It is not. Not knowing is the starting condition of every piece of knowledge anyone has ever produced.

Sit with it. The discomfort will pass. A doubt that has been sat with long enough turns into either a resolution (you have found the answer) or a more precise question, which is itself progress. Doubts about why a friend is sulking are usually settled in an hour. Doubts about dark matter (the unseen stuff that seems to hold galaxies together) have taken nearly a hundred years and are still being worked on. Some questions get answered in a single afternoon. Some take generations. The size of the question decides the size of the work.

The last route I want to walk you through is the most carefully built one of all. It is the one science is mostly travelling on. The scientific method, when you strip it of jargon (the special words of experts), is a particular kind of checking. It is slow, repeatable, and designed to give the same answer regardless of who is doing the checking. It is not the only road to knowledge. It is the most reliable one we have for the parts of the world it can reach.

I want to say one thing before I walk you through it formally. It is that this kind of checking is not something only scientists do. A shopkeeper who notices that the price of the tyres he is buying from his usual supplier has gone up uses the same procedure. He observes the change. He asks why. He guesses at a reason. Maybe the supplier is gouging him, charging more simply because he can. Maybe the cost of rubber is up worldwide. He tests his guess by calling around, comparing other suppliers' prices, watching the trade reports. He revises his picture. He decides what to do: switch suppliers, eat the higher cost himself, or raise his own prices. He has done five-sixths of what a chemist does in a lab, in the back room of his shop, without ever using the word "experiment." Most of the useful thinking ordinary people do is some version of this. The formal scientific method is just the same procedure with more rigour (more care and strictness) and more people checking each other's work.

Now the formal walkthrough, with a small example. Imagine you sit down on the sofa, point the remote at the television, press the button, and nothing happens. The screen stays black.

You start with what you noticed. The television did not come on. That is your observation, even if you do not call it that.

You ask yourself why. That is the first move that separates a thinking person from somebody who just shrugs and goes to bed. Asking the question is what turns observation into work.

Then you guess. Maybe the batteries are dead. The guess might be right. It might be wrong. What matters is that it is a guess of a particular kind: one you can actually check. If your guess had been "the gods are angry with my television," there would be nothing you could do to confirm or deny it. There would be no way for the world to talk back. The technical word for what separates a real guess from a wish is falsifiability. It is just the long word for this: imaginable evidence that, if it came in, would force you to give the guess up. Without that, your guess is not really a guess about the world. It is a feeling.

From the guess, you draw a prediction. If the batteries are the problem, then changing them should make the remote work. Notice that the prediction is specific. "Things will get better" is not a prediction. "The remote will work after I change the batteries" is one.

Then you do the test. You change the batteries. You point. You press.

Two things can happen. The television comes on, in which case your guess was probably right, and you have learned something. Or the television stays black. If it stays black, two further things are possible. Either your guess was wrong (the batteries were not the problem), or your test was bad (the new batteries you put in were also dead, by bad luck). So you try yet another set. If nothing works, you go back and propose a new guess. Maybe the sensor on the remote is broken. Maybe the television itself is dead. Maybe the power has gone out in the building. You test the new guess. You watch what happens. You keep going.

This loop is not a one-time procedure. It is a feedback machine that keeps running until the question is answered. And even when you think it is answered, the conclusion is still open to revision. Tomorrow, somebody might find a better explanation for what you had concluded was settled. That is not weakness. That is the only way knowledge has ever grown. The willingness to be wrong is the engine. The loop stopping is what we call dogma. Dogma is a conclusion that is no longer allowed to be checked.

I am walking you through this slowly because of what happened when I finally understood it as a procedure I could use, not a thing scientists did somewhere far away. It changed how I read everything. Every news story. Every claim from somebody in authority. Every old answer I had been carrying around inside me. The question I started asking, of any claim that mattered, was a simple one: what would force you to give this up? If the answer was "nothing could," I stopped trusting the claim. A claim that nothing could falsify is not strong. It is invulnerable: no evidence can ever hurt it. The two are not the same thing. Invulnerable is exactly what a feeling looks like when it is being passed off as a fact.

Some hypotheses are bigger than batteries. Sometimes a hypothesis is about how the world works at a level you cannot put on a kitchen table. Then you have to be more careful about what could be confusing your results. The way scientists handle this is the controlled experiment.

Say you want to find out whether doing yoga every day for six weeks actually improves people's health. The lazy way is to find somebody who does yoga, watch them for six weeks, and conclude that any change in their health was because of the yoga. This conclusion would be unreliable. The person might have changed their food. The person might have changed their sleep. The weather might have improved. Too many things are changing at once for any one of them to be safely picked out as the cause.

So you do something more careful. You take fifty people. You check their health at the start. You split them into two equal groups. One group does yoga for six weeks. The other group does not. Everything else about their lives (what they eat, when they sleep, where they live) is kept as close to identical as possible. After six weeks, you check the health of both groups again. If the yoga group has improved and the other group has not, you can say with some confidence that it was the yoga. If both improved equally, the yoga was not the cause. If neither improved, the practice did not work.

The group doing the thing you are testing is called the experimental group. The other group is called the control group. The thing you are deliberately changing, in this case whether the people do yoga, is the independent variable. The thing you are measuring, the change in health, is the dependent variable. The technical words are not what matters. The shape of the test is what matters. One thing changing, everything else held still, and a comparison group to show what would have happened without the change.

I have run informal versions of this kind of test on myself, on smaller questions, many times. Does this food make me feel worse the next morning? Does sleeping at this hour, instead of that one, change how the day starts? The hardest part (the part I am still bad at) is holding everything else still. Real life is messy. You change the food but you also have a stressful week. You change the sleep but you also start a new exercise. You attribute the result to one thing when several things shifted at once. This is why scientists go to such lengths to control everything but the one variable they care about. They know how easily a busy life fools an honest observer. You will know it too, if you try.

Most things in real life are not as clean as a yoga study. You cannot always run a controlled experiment. You cannot ethically test on people whether a poison kills them. You cannot rerun the weather. So scientists settle for the best they can do, which is often careful observation of natural variations. They look at what happens to one population versus another over time, and at what happens before and after some event the world ran for them. The conclusions are weaker than from a clean experiment. The honest scientists say so. The dishonest ones do not.

One more thing about experiments. A small sample (the group of people you test) is rarely enough. Fifty people is a start. Five hundred is better. Five thousand is better still. The bigger the test, the smaller the role chance plays in the result. This is also why the same experiment has to be run by different people in different places before its result is considered solid. One result is a hint. A hundred results, from a hundred labs, pointing the same way, is knowledge.

I want to end this section with a word about mathematics. I have a complicated feeling about it, which is worth being honest about. I respect what mathematics can do. I have come to be suspicious of what it is sometimes used to do. Both things deserve saying.

Mathematics is, in some ways, the language the universe speaks to us. It lets you say not "the apple falls down" but "the apple accelerates at nine point eight metres per second squared." It lets us be exact where ordinary words leave us vague. Many of the most reliable pieces of knowledge our species has produced are mathematical in form. Everyone should learn enough of it to use it.

But mathematics has limits, and they are worth knowing. The first limit is that mathematics is not a substitute for the world. A mathematical proof of something is not a proof that the something exists in the world. You can prove all kinds of things on paper that have no counterpart out here, nothing in the world that matches them. People have used elegant mathematics to argue for parallel universes, for invisible dimensions, for whole catalogues of unobservable things. The mathematics works on the page. There is no reason yet to believe any of these things exist out in the world. A mathematical argument for an unobservable thing is only an argument for what could be. It is not a proof of what is. The test still has to come from outside the page. We should not let ourselves be dazzled by elegant equations into accepting whole imaginary worlds the equations cannot reach.

The second limit is more practical. Mathematics works extremely well for things that behave consistently: atoms, planets, currents, large populations averaged out over time. It works extremely badly for one specific kind of thing: the behaviour of one human being. People do not act like atoms. They act according to histories the equations cannot see.

A stock might go down because the company is failing. Or it might go down because thousands of investors all needed cash this week for personal reasons that have nothing to do with the company. One man may sell his good shares to pay for his daughter's wedding. Another sells to pay a medical bill. A third sells to start a small business. The equation cannot see the wedding or the hospital or the dream. The equation will give you an answer. The answer will sometimes be very wrong. It is, in the end, idiotic to think you can predict everything human beings will do through equations. Use mathematics where it actually works. Be modest about it where it does not.

So here we are, at the end of a long walk through the routes of knowing. The ones in this chapter are the main ones. You will use most of them daily, whether or not you have thought about them this way. The question is what you do with what they hand you. Do you ask, before you lean on it, whether it will hold? Most people do not. They lean on what they were given, by people who themselves were given it by other people, going back as far as memory reaches. And they call this leaning common sense. Sometimes the leaning holds. Often, it does not, and they fall. They fall slowly, over years, through bad relationships and bad jobs and bad choices, without ever being able to name what gave way under them.

What I want you to feel (not just understand, but feel in your chest) is that knowing is not a passive thing that happens to you. It is something you do. You can do it well. You can do it badly. You can also refuse to do it at all, and let other people do it for you, and let them tell you what to believe at the end. That is the most common option. It is not a good option. It is the option that has produced most of the unnecessary suffering in human history. The world has many confident-sounding people who will be glad to do the checking for you. And they will be glad to charge you something for the service: money, devotion, obedience, your only afternoon. They have not actually done the checking. They have only got better at sounding as if they have.

The cost of doing your own checking is that it is slow and uncomfortable. It sometimes turns up that things you had loved for years were never true. The benefit is that what you end up with is yours. Nobody can take it from you. It will hold your weight. It will not unravel when you change cities, change beliefs, or grow old. It is the only kind of knowledge that holds up across a whole life.

There is one more thing I had to learn before any of this was much use to me. Having the tools is not the same as using them well. A person can know all about hypotheses and testimony and the scientific method, and still reason badly. The mind has its own bad habits. There are traps that even careful thinkers fall into, again and again, because the traps are older than thinking. The mind walks into them before any conscious checking can begin. The next thing I had to figure out was what separated a person who uses these tools well from a person who has them and still gets fooled.

That was the next stretch of road.

Words to Keep

• knowledge: what is still standing after you have leaned on it. The shared body of theories that have survived honest checking.

• belief: something a person has decided is true without ever testing it. A starting point, not an enemy, until you mistake it for knowledge.

• hypothesis: a belief stated clearly enough that someone could go out and check it, and could find it wrong.

• falsifiability: the room a claim leaves for some imaginable evidence that, if it came in, would force you to give the claim up.

• theory: several tested hypotheses locked together into one coherent description. The strong end of the chain, never just a guess.

• fact: a short, usable piece of knowledge that you can put to work without following the whole chain that produced it.

• rationalising: putting reasons around a belief you have already decided to keep, instead of putting it in front of a test that could destroy it.

• the four pieces of knowing: the observer, the idea, the result and the means of checking. Take one away and you only have something that looks like knowing.

• direct knowing: what your own senses bring you. Fast and useful, but it fails through wrong perception, faulty senses and imagined things.

• inference: walking in the mind from what you can see to what you cannot, as from smoke to fire, along an assumed connection.

• testimony: knowledge passed from one person to another. It works only when the teller got it right and says it clearly.

• controlled experiment: a test in which one thing changes, everything else is held still, and a control group shows what would have happened anyway.

Sources and Further Reading

• Gautama, Nyāya Sūtra. The classical Indian account of the means of knowing. The five-step inference used here (fact, reason, example, application, conclusion) and the smoke-and-fire example come from this tradition.

• Karl Popper, The Logic of Scientific Discovery. The source of falsifiability, the test this chapter uses to separate a real hypothesis from a feeling.

• Edmund Gettier, Is Justified True Belief Knowledge? (1963). The short paper that unsettled the textbook definition of knowledge I argue with in this chapter.

• Albert Einstein, Relativity: The Special and the General Theory. Einstein's own plain account of the theory whose light-bending prediction this chapter uses as its example of a hypothesis promoted.

• Frank Dyson, Arthur Eddington and Charles Davidson, A Determination of the Deflection of Light by the Sun's Gravitational Field, from Observations Made at the Total Eclipse of May 29, 1919 (1920). The report of the eclipse test in which the light did bend.

• Irving Wallace, The Fan Club (1974). The novel that the story of Sharon Fields and Adam Malone comes from.

• Carl Linnaeus, Systema Naturae (1735). Where the system of ranked groupings that my cyclist relies on began. Later scientists added ranks such as phylum.

• Caraka Saṃhitā. One of the classical texts of Āyurveda, my example of the holistic way of healing.

• Nithin Palal, The Philosophy of Everything, first edition (2022). The original book, where the Sharon Fields example and much of this chapter first appeared.