How We Know What We Know

关于科学：科学家怎么把一个好奇的问题，变成别人也能相信的结论——以及这条路走到哪里就到了尽头。

The Boy Who Kept Asking

Let me start with a boy named Ken. He was nine years old, small for his age, and never quiet. His mouth was usually open, and it was open because he had another question. Why is the sky blue? Why do grown-ups say one thing on Monday and do the opposite on Friday? Why does the moon seem to follow the car down a dark street?

Ken was not trying to be clever. He was curious, and curiosity is the spirit behind all research. He did not know that the word "science" grows out of an old root meaning "to know." He only knew that not knowing felt like an itch he had to scratch. When you are young, every question is enormous, and a question is a powerful thing even when it sounds silly to a busy parent.

His neighbor, an old engineer named Sato, gave him advice that stayed with him for forty years. "Cut your question small," Sato said, "and it will either answer you or break you." Ken did not understand this at all. It was one of the strangest things an adult had ever said to him, and he thought about it for years. The idea underlying all modern science is simple enough to say and hard enough to live by: a question is not a wall. It is a door, and every door has a handle.

Ken normally lived inside the quiet of that domestic world — school, dinner, homework, sleep — and he was not unhappy there. He was glad when the water boiled. He was comfortable with the house. But he did not rely on it to answer anything, and he did not depend on being told what to think.

Once a week he would participate in a school sport, badly, at the back of the team, where nobody really looks. The best player on the school, a girl called Ana, could hit the ball so hard that everyone swung back to swing at once and the whole class shouted. Ken liked the noise more than the game. It was the first time in his life he had noticed that a physical motion could be studied instead of merely watched.

He was not active in the way that word is usually used. He was quiet, and he was glad to be left alone with a box of things. But he was entirely unable to leave a question alone, and other people could sense it. Sato once told him that this was a fancy word for the same thing: a person who designs his own trouble and is then frightened when the trouble arrives.

That is not a small problem. A boy who cannot stop asking has no rest, and no rest affects everything — sleep, friends, the way a teacher treats him. Sato's advice was besides the science: learn to fold the question up and put it away until Saturday. Ken could not do this at first. Later he got extremely good at it, and the questions never went away. They just sat quietly in a drawer, bright and sharp, until he was ready.

Cutting a Question Small

A large question is hard to test. "Is the moon alive?" is enormous and useless. But "Does the shadow of the moon grow longer when the light sits lower?" is small enough to answer on a kitchen table with a ball, a lamp, and a ruler.

A hypothesis is a possible answer that you can be wrong about. That last part is the entire point. If you cannot be wrong, you are not doing science; you are describing a belief. People once held very romantic ideas about nature. The sky was a great glass dome. A storm was the anger of a god. The stars were holes in a roof. Some of those ideas were beautiful, and the people who held them were not stupid. They simply could not be checked, so they had to be abandoned — not because they were unkind, but because they were untestable. A theory that hides from every possible test is not a strong theory. It is a wall.

To test an idea, first make it exact. Scientists call the careful version a principle, and a good principle sounds dull on purpose. It tells you what will happen, in what direction, and by about how much. "Hot air rises" is a principle. "The moon is beautiful" is not a principle; it is a feeling, and feelings are not wrong. They are just not evidence.

You also have to decide what you aim at before you start, because a measurement you did not plan is usually a measurement you cannot use. The trick of a good question is that anyone can repeat it and get the same answer, or at least close to it. Ken used to say his questions were good when his mother could answer them. He was wrong. A question is better when the world can answer it.

Then go and find the variable. In every experiment there are three players. One thing you change on purpose. One thing you hold steady on purpose. And one thing you simply watch. If you change the temperature, temperature is the variable. If you keep the quantity of water the same, that is a constant. And the second cup, untouched, sitting right beside it, is the control. Without a control you have nothing to compare, and comparison is the whole job.

You also have to be ready to fail. Most school experiments fail, and that is fine. A failure with a clear reason is worth ten successes you cannot explain. And you must be ready to refer to other people's work, because nobody relying on their own memory of a hundred years of results is going very far.

In the Laboratory

A laboratory is a quiet room full of equipment. A beaker. A balance. A bottle of clear liquid with a label in small careful letters. A microscope with one cell in the light. A long cable strung across the bench. A box of instruments that cost more than a car, bought from a retail shop by its owner, then carried up two flights of stairs by two people who are not paid enough.

The work is usually duller than you hope. A sample goes in. A number comes out. The number is written on a line, and then the same measurement is made again, because the first one might be wrong. Ken learned this from his friend's older assistant, Reiko, who had been doing it for eleven years. "Nobody wins a prize for being right once," she said. "Being precise is a habit. You are exact in the morning, and you are exact again after lunch, when you are tired and would rather go home." She said this herself, one afternoon, and then assured him, wrongly, that she did not mind his questions. He appreciated it for years without knowing why.

If the number changes when nothing about the experiment has changed, you have found a variable you did not know you had, and you must hunt it down before you can trust anything. This is why labs are kept clean, dim, cool and quiet. The constant is not the room. The constant is everything you failed to think of. Forgetting to control for one small thing can complicate a result for twenty years.

Meanwhile, ideas in biology were moving faster than any equipment could follow. A gene, said to be a recipe. A protein, said to be a small machine that builds and repairs. A chemical code inside every cell, so genetic information can be copied, changed and passed on. A tiny infection that moves from one cell to another, far too small for the old textbooks, visible only through glass and a good lens. Nobody could look at a gene in 1950. The tools did not exist yet. Science was simply waiting for the right instrument, the way a language waits for the right verb.

Keeping score.

Science keeps score in a very plain way. Everything becomes a number, and the numbers are allowed to be ugly. A statistic does not care about your feelings.

Say you want to know the effect of light on plant growth. You set out forty plants, twenty in light and twenty in dark. Every week you measure the height. After eight weeks you calculate the difference, and the difference turns out to be small. Small is a real answer. Most real answers are small.

Sometimes a result is too small to be real. Then you have to be honest about it, and in statistics this mistake even has a name: the null result. It does not mean that nothing happened. It means "I looked hard and did not find it." That is still a conclusion, and a good one, because it tells the next person where not to dig.

Trouble begins when somebody explains too much. If your data is thin, you can attribute almost anything to it. Watch for a speaker with twenty samples and a very confident claim. It is either very good or very wrong.

The other bad habit is the negative result that gets quietly thrown away. A scientist has a tendency to publish the exciting finding and forget the boring one, and if nobody repeats the boring one, the gap in the record is permanent. Good labs keep their failures on a shelf, not in a drawer.

A clean analysis takes time, and time is the one thing nobody can supply. Reiko used to say that the last hour of a study is worth three, because that is the hour where you finally notice what you have been blind to all week. The word "notice" is important. Every good experiment begins with observation, and observation is a skill, not a personality. Careful people see more, because they are ready to see.

You analyze the data twice. Once to find the pattern, and once to try to break the pattern you found. Most scientists are, at heart, extremely smart people who spend their lives trying to prove that they are wrong. The skill is not in finding the answer. It is in knowing, the instant the answer appears, how it could still be a fault in your own work.

Sometimes the honest thing is to extract only a little from a lot of work. Ken's first real discovery was that a paper cup does not leak faster when it is hot, which he had assumed for years and which turned out to be true for a reason he could situate only after three weeks. Small, dull, useful. He agreed with himself that this was the whole job.

What the Body Knows, and What It Cannot

The body was the first laboratory, and it is still the least honest one. A muscle does not care what you believe. A nerve does not take your side.

Everything you know about your own body arrives through your perception, and perception is a creative editor. It fills in the half-second when your foot is asleep. It makes the moon look bigger than it is. It turns a noise into a word if it expects a word. Which means that the most interesting tool a scientist owns is not a telescope or a balance. It is the ability to avoid the pattern his own mind wants to impose, and to listen to what is actually there.

There are hard limits here too. Your senses only reach so far. Instruments exist precisely because the eye is a terrible witness and cannot see a cell, a gene, or a billionth of a second. That is not a weakness of the body. It is a capability: it means a person can invent a tool, and then extend what a person can know. Every instrument begins as somebody's very creative refusal to accept the first answer.

The body also consumes energy at an alarming rate, all day, without being asked. It injures itself easily under load, heals itself constantly without being thanked, and will exceed its own limits if you ask it to. The margin between those two edges is narrower than people think.

The shape of an argument.

It helps to situate a scientific claim in ordinary speech before you believe it. In the grammar of everyday language, a claim has a subject and a verb, and the verb is where the risk sits. "Plants grow faster in the dark" is a claim you can test. "Some plants are sensitive" is a phrase you can only nod at. One is a sentence with a door in it. The other is comfortable furniture.

Good scientific writing has a variety of sentences, and it should. A paper where every paragraph says the same thing with different words is not clear, it is thin. The reader needs a short sentence for the hard idea, and a long one for the messy data, and he should be able to feel the switch.

Watch for the person who uses a suit of careful words where ten plain ones would do, who makes every limitation sound like a finding. That is not always dishonest, and it is usually invisible. The cure is unglamorous: ask for the number, ask for the extent of the claim, ask how far it was tested, and touch the soft parts until something moves.

One last warning, since it costs almost nothing and saves a lot. Two things can influence you without your noticing: a compensation paid to whoever reports a result, and a habit of only reading papers by people whose names you already recognize. The first is a matter of law and money. The second is a matter of taste, and it is more common, and it is what keeps a young field small and quiet for twenty years.

A small honest experiment.

Ken ran one that nobody in the class could do better. He netted three different leaves, weighed them on the same balance at the same hour, and wrote the numbers down without rounding. The point was not the answer. The point was that the last digit was still moving like a natural thing, slightly, day by day, and the only way to see that was to record it rather than guess at it.

Because it matters

Because it matters as much to know how a thing gets conducted as how it turns out, here is a small test you could actually run.

Put a paper cup and a glass cup on the same table in the same room. Weigh both of them at the same hour for a week. Pour the same exact amount of water into each. Then leave them alone. Do not neglect them, do not move them to a window, do not touch the table.

Most people will predict one result. Almost all of them will be ideal about it. And in a careful run, the difference between the two cups over a week is tiny — not zero, because a paper cup breathes, but tiny, and much smaller than the difference between day one and day seven. The result is dull, and it is accurate in a way that a dramatic result rarely is. A flag you can hold up: an experiment whose answer is small and boring is usually an experiment that was conducted honestly.

The reason to actually do this is not the answer. It is the four days. It is that for four days you contribute three minutes to something that will outlast you, and nothing in ordinary life asks that of you.

The Museum Behind the Glass

The best classroom in the world is a room with a type of nothing in it at all: no answer, no summary, only the thing itself. Ken's school had a small museum, and the entrance to it was a heavy door that opened onto a room so cold that visitors complained.

In the first case lay the neck of something enormous, and the label said the name of a scientist who had dug it out of soft southern rock. Beside it was a wing with holes in it, so thin that light came through. On the wall, an illustration of the animal as it must have looked when it was alive, drawn by somebody who had never seen one, using only bones and a lot of guesswork. Ken stared at that picture for twenty minutes and felt something move in his chest that he could not name. It was the moment science stopped being a subject and became a place he wanted to live. He was frightened by how little he knew, and he was glad of it.

The deposit came next: stone, cut open, with a thin dark line parallel to its surface. Schoolchildren were told that a mine nearby had found the same layer at a slightly different angle. When two rocks, found far apart, tell the same story, the story unites into something larger than either rock, and that is not magic. It is just perspective. You stand far enough back and the pattern appears. Ken never forgot the feeling of that. The world had not changed. He had only acquired a new way to stand in it.

A panel of teachers met once a term and criticized the museum, which was liberal in its funding and lucky in its space. One of them, an academic who taught mathematics, argued heavily that the labels should be shorter. He won. It was the only exception to the general rule that the museum went unvisited in August.

Two hundred years ago all of this was legal to do and completely illegal to publish. There were laws about what a person was allowed to say about the earth, and the penalty for breaking them was not a fine but a prison. When a requirement like that is removed, it is usually called an amendment, and the whole content of science changes in a single generation. The chief thing to understand is that science did not win an argument with the church. It won because someone finally built an instrument that gave the same answer twice, in front of a sober man who had every reason to hate the answer.

The Space on Top of Us

Then there is the space, which behaves in a way nobody expected.

A long time ago people watched the sun, the moon, the stars and the planets, and told stories. The stories were good. They explained the harvest, the fright of a dark month, the grief of a person standing at the edge of the sea. The story was not a lie, but the star in the story was not made of burning gas, and the sky did not sit on the shoulders of a tired giant.

The surprise was how precise the numbers turned out to be. A planet does not wander, it travels, and it does it at the same speed for every observer on the ground. The moon does not follow your car; it is simply very far away, and your car is moving too. A comprehensive study of a thousand years of records found a spot where an unseen body pulls, and the pull had been sitting in plain sight the whole time.

The proof was arithmetic, not argument. That was the real breakthrough, and it took a while to sink in. Before, a correct answer was something you could argue towards. After, it was something you could substitute into a line and check. You could take the number of days in a year and conduct the sum yourself, with a string and a stick and a pile of pebbles, and arrive at the same number they did. That is an impressive kind of power, and it arrived all at once.

A serious argument about the sky is like a map. It is not a copy of the whole country, but it is a guide you can walk with. A map is a reflection of the territory. The map is not the territory. But if your map is bad, you will fall off the edge of the world.

For a long time the map had an enormous hole in the middle. Then a man with a good telescope and a very patient nature worked out a formula, and the equation for gravity came out of his head, and then out of a page, and then into every schoolbook on earth. Newton knew almost nothing about the motion of bodies in space. He worked it out on the ground, with an apple.

Here is the strange part: every one of those words is a label for something you can observe. You cannot see a force. You can see a sheet of paper fall. You cannot see mass. You can see the bottom of a boat sink. Science is the habit of pointing at the word, and then pointing at the world.

The electrical force, the pull between two pieces of charged dust, is equivalent to the pull between two pieces of mass. You cannot feel one without feeling the other, because they are the same thing wearing different clothes. A drop of water at the top of a hill has more energy in it than the same drop at the bottom. A wave of heat spreads from hot to cold. A current of electricity moves through a wire. All of it is motion, and all of it obeys a fundamental set of rules.

Where the money and the politics arise.

Nobody does science alone, and almost nobody does it for free. A laboratory needs funds, a salary, a visa, a permit. Which means a scientist must eventually deal with institutions, and institutions have their own weather.

The official version of this is dull and mostly honest: your registration is checked, your results are recorded, your paper goes out, somebody reads it, somebody criticizes it. The unofficial version is worse and more immediate. Somebody makes a phone call. Somebody decides your name is not on the list this year. You discover that the young assistant who did all the actual work is not on the paper at all, and by the time you argue, the paper is out.

There are substantial reasons these small things happen. Labs are short of funds. Senior people are judged on how many papers they produce. Journals prefer results that are clean over results that are true. And nobody gets promoted for telling the public that a promising result did not hold up, though everybody cares about it very much when it is their own health.

None of this means science is fake. It means science is made by people, and people are a messy and occasionally a rotten lot. The defense was built a long time ago and it still works: open data, real peer review, and the quiet stubborn fact that a result which somebody tried to bury usually survives, because somebody else with better equipment will notice the question. It succeeds in the end, not because everybody is good, but because the system is built so that one honest person is enough.

And when the result is big enough to matter — a new material, a new drug, a new invention — the money arrives immediately, and with it every kind of pressure a person can be under. This is the immediate danger and the slow danger at once.

A Paper Is a Promise

And then there is the part everybody forgets, which is the slow part. Science does not end at your desk. It ends in a paper.

A research group may spend a year on the experiment and six more months writing it up, because a result that is not written down carefully will not last. They must explain the method well enough that a stranger in another country can repeat it. They must state the result in plain words, display every unit they used, draw one clean figure, and then pay somebody to check every detail twice.

When the paper is printed in a journal, anybody can read it. Anybody can walk into a lab and try to break it. That is the deal. A journal does not publish a paper because it is true. It publishes it because somebody was willing to be wrong in public and gave others the tools to find the error.

A result becomes reliable only when other people repeat it. Not once. Twice. Three times, in three different regions, by people with no stake in the answer, on equipment that looks nothing like theirs. A single demonstration is a performance. Three demonstrations in three countries are knowledge.

This is also why a conference can be so strange. For four days, two hundred people who have spent their lives alone with a problem gather in one hot room and argue about it. Nothing is decided. Nothing is won. At the end, roughly one idea out of the two hundred is ready to be tested, and the person who proposed it goes home and does the work. The reward is not the room. It is the excitement of being slightly less wrong than you were in January.

Where the map ends.

And there are places where the map stops.

We do not know how the universe began. We have several very good theories, and the biggest gap is not the math. The biggest gap is that we have never observed the first moment. We do not know what the brain is doing while it decides. We do not know why the universe is made of the particles it is made of and not of some other set. And there are good arguments that these are not temporary gaps. They may be permanent.

Sometimes the honest answer is "we do not know," and saying that is a form of strength, not weakness. It is exactly what Ken learned at fifteen, when his theory that the children in his class were all secretly robots fell apart in a week. He was ashamed for a day, and then he noticed that his friend Maya had written down the same idea two days earlier and had it killed by the same sentence. They laughed. Then they went and touched the doorbell, to find out which kind it really was.

Limits are not the opposite of science. Limits are where science begins. The researcher who is aware of what she does not know is more useful than the one who is not. So is the friend who is not afraid to say "that is a bad argument," as long as she also says what would make it good.

What It Is Like Now

Look at what has developed since Ken was nine.

A satellite the size of a fridge sends his question about the sky straight to another planet, and gets an answer in a session that lasts nine minutes. A laboratory runs itself overnight, and a person is awake on the other end of it, three time zones away, who has never met him and never will. At a meeting in the eastern part of Germany, a hundred people argue for six days about a number with nine digits after the point. None of them will win. The official record will belong to the number, not to the person.

Yeah, it is crazy. But here is what did not change: someone still has to design the experiment, someone still has to sit at the bench and mark the trail of the measurement carefully enough that another person can follow it, and someone still has to neglect their own career for a while to do it. The instruments got impressive. The craft did not change at all.

A doctor in the same building can look at a drop of blood and derive a story about your liver. A lawyer can use the same story against you. A company can suspend a product on the strength of it. Somewhere between here and the moon, an engineer is solving an equation that was written down by a woman who died before she was forty, and her name is the reason he is alive.

And in a school science fair somewhere, a kid is standing behind a table on a small stage, holding a piece of paper and a cup of water, saying the word "hypothesis" in a slightly too loud voice. Her voice is soft, then it is not. Two hundred other children are bore by the whole thing, and one is trying hard not to stare at his shoes. Somewhere at the back, a girl whose project was stolen last year by a boy who copied her idea and won is doing absolutely nothing about it, because rules are rules and the chief judge has already written her name down. She will find out later who tried to steal it, and how, and it will reflect very well on her.

She has five minutes. She has a control, a variable, and a result that is not what she expected. She says "the outcome was negative," and three people in the front row sit up. Her hands are cold. Her teacher, who was the holder of the school record for the youngest winner ever, is watching from the side with his arms folded.

What happens next is not luck and it is not talent. She takes a breath, and she tells the truth about what her data does and does not show, and the room changes temperature. That is the whole trick. That is the entire, ultimate skill. Everything else is just equipment.

The universe is under no obligation to make sense to you. But the part of it that does make sense — the moon, the water, the heat, the math of a swing, the reflection of light on glass, the enormous quiet sky — all of it is open to anyone who is willing to look, measure, and be wrong in public.

Science is one long article that the whole species keeps writing, and every innovation in it is a line somebody added without ever getting credit. None of it rests on nature being friendly or on any person being reasonable. It rests on one enormous assumption: that the world will keep behaving the same way till tomorrow, and that you can escape back out of your own assumptions when they go wrong. Break that second half and every measurement becomes meaningless at once. Nobody has managed it, automatically, and several people have tried, usually with good funding.

You do not need a degree. You need a question, a ruler, and a friend. Ken had a ruler. Sato had the door. The rest is only practice, and the habit of practice is what underlies everything a laboratory has ever built.

If you have a friend who asks terrible questions, the kind that make a grown-up go quiet, give that friend a candle and a dark room and then explain nothing at all. Just look up with them. That is the oldest experiment there is. It still works. And you are extremely lucky to be alive on a planet where it can be done at all.

生词与提示