What is science?

Someone records a video with their phone to explain that the Earth is flat and that gravity is a fairy tale. Upon finishing, the device adds the exact location. It calculates it by listening to a constellation of satellites orbiting a spherical planet, held by the same gravity that the author of the video just denied. The clocks on those satellites are adjusted to compensate for 38 microseconds a day, just as Einstein's equations require. The video is published in seconds. That night, thousands of people will see it.

We have never known so much. In 2023, the Scopus database alone registered 3.3 million new science and engineering articles, more than six per minute. It is estimated that vaccines have prevented 154 million deaths since 1974. And it has never been easier to deny everything without getting off the couch.

Measles, for which there has been a vaccine since 1963, infected about 11 million people in 2024 and killed about 95,000, most of them children under five. The vast majority of these deaths occur in poor countries or countries with fragile health systems, where the vaccine does not reach everyone. That year, there were large outbreaks in 59 countries, almost triple the number from three years earlier.

But the virus is also returning to places where there is a surplus of vaccines, to countries that had eliminated it—that is, where it had stopped circulating continuously. The United States achieved this in 2000. It closed 2025 with 2,289 cases, its worst figure since 1991, and by early September 2026, it already totaled 3,134, nearly nine out of ten in unvaccinated individuals or those with unknown vaccination status. In November 2025, the American continent lost its status as a region free of endemic measles after the virus circulated for at least one year straight through Canada, especially in communities with low vaccination rates. In January 2026, six European countries lost that same status, including the United Kingdom and Spain, which had held it since 2016.

Other diseases have already followed that path. In the last decades of the 20th century, campaigns against the pertussis vaccine caused vaccination rates to drop in several countries, including the United Kingdom and Japan, and the incidence of the disease there reached ten to one hundred times higher than in countries that continued to vaccinate. In 2003, three states in northern Nigeria suspended polio vaccination due to rumors that the doses contained sterilizing hormones and the AIDS virus. The boycott lasted eleven months. The virus reinfected eight African countries that were already polio-free and jumped to Asia, where it left more than 1,500 children paralyzed.

In a 2021 online survey of 1,134 adults in the United States, one in ten agreed that the Earth is flat, and nearly as many said they were not sure. An analysis of 126,000 verified stories that circulated on Twitter between 2006 and 2017 found that false ones were 70% more likely to be shared than true ones.

The majority still trusts. When an international team asked 71,922 people from 68 countries, the average trust in scientists turned out to be moderately high, and no country fell into the low range. But that trust is eroding. In the United States, where it is measured almost every year, it fell from 87% in 2020 to 73% in 2023, and in 2025 it remained ten points below the starting point. The conviction that childhood vaccines are important declined during the pandemic in 52 of the 55 countries analyzed by the United Nations Children's Fund (UNICEF).

Never before had so many people heard science called into question. Today, anyone with a phone can reach millions of people without anyone asking them for a single shred of evidence. Part of that misinformation is a business, and the rest is ignorance with a loudspeaker. Those who are distrustful remain a minority, but measles is easily satisfied. Curbing its outbreaks requires that at least 95% of the population have the two vaccine doses. The virus doesn't care if the rest are missing due to refusal or because the vaccine never arrived.

The question of how we know something is true is much older than phones. Throughout almost all of history, the usual answer was to obey those in charge and custom. There were always those who preferred to go and see for themselves, but that other answer, which is much more uncomfortable, only began to take hold a few centuries ago. It consists of measuring and letting others measure too, especially if you dislike them.

It is therefore advisable to be clear about exactly what it is that some defend without having looked at it closely, and others attack without knowing it.

What science is and how it knows what it knows

Science is a method for finding out how the world works and, at the same time, the body of knowledge that this method has been accumulating. It observes, proposes an explanation, deduces what should happen if that explanation were true, and goes to check it. What passes the test is kept as long as it continues to pass tests. What fails again and again ends up being discarded, no matter who said it.

By definition, science is an ordered body of knowledge obtained through observation and reasoning, from which laws that predict and can be tested with experiments are derived. The word comes from the Latin scientia, knowledge. The weight of the definition is at the end, in predicting and testing, two requirements that almost no other field of knowledge imposes on itself.

The scientific method, a circuit with feedback

The scientific method is usually taught in six steps. Observe, ask, propose a hypothesis, deduce a prediction, put it to the test, and publish the result so others can repeat it. The six form a circuit. When the prediction fails, you go back and revise the hypothesis. When it succeeds, someone tries to tear it down from another angle.

More than 2,200 years ago, Eratosthenes, a librarian in Alexandria, knew that in Syene, modern-day Aswan, objects did not cast a shadow at noon on the summer solstice, while in Alexandria they did. He assumed that the Earth was a sphere and that the Sun's rays arrived parallel, measured the shadow, took the distance between the two cities, and calculated the circumference of the planet. He got 250,000 stadia. Historians still argue about how long his stadium was and, therefore, how wrong he was. The procedure, however, can be repeated today by a school with two sticks and a phone call.

Not all science fits in a laboratory. An astronomer cannot turn on a star to see what happens, and a paleontologist cannot repeat the Cretaceous. The method works for them just the same. They deduce what should be observed if their idea is correct and go look for it, with the real risk of not finding it. That risk separates a scientific hypothesis from an opinion.

Hypothesis, law, and theory, three words that get confused

These three words change meaning when entering a laboratory. A hypothesis is a proposal that still needs to be tested. A law describes a regularity, for example how an object falls, without explaining why it happens. A theory is the explanation. Someone on the street who says "it's just a theory" is using the word that in science names a tested explanation as a synonym for a hunch. The National Academy of Sciences of the United States defines it as a well-substantiated explanation of some aspect of the natural world, capable of incorporating facts, laws, inferences, and tested hypotheses.

A theory does not become a law when confirmed, because they answer different questions. Gravitation has laws that describe falling and a theory, general relativity, that explains it. The theories that have withstood the most tests are among the most solid things we have. The germ theory, the atomic theory, plate tectonics, and evolution have been passing them for decades. We work with them daily, from the operating room to the seismic observatory.

A self-correcting machine

Scientists make mistakes and sometimes lie, like anyone else. The system counts on it. A result goes through review by specialists outside the study before being published, and if the matter is important, other teams try to repeat it. In 2015, 270 researchers repeated one hundred already-published psychology studies. 97% of the original ones had yielded a statistically significant result. Among the repetitions, 36%. It was a blow to the discipline, and it dealt it to itself in public, something few institutions do of their own volition.

The myth linking vaccines and autism shows how that machine works and how long it takes. In 1998, an article based on twelve children suggested a link between autism and the MMR vaccine, which includes the measles one. Other teams went to look for it and didn't find it. The journal retracted the article in 2010 after the British General Medical Council deemed it proven that it contained false claims about how the children were chosen and about ethical permissions. A year later, the medical journal BMJ published the investigation by journalist Brian Deer and labeled the work a fraud. In 2019, a Danish study went back to look for the link in 657,461 children, nearly 55,000 times more than in the original, and also found no increased risk of autism in the vaccinated. The myth still circulates. The correction, which is the scientific part of the story, travels much slower. Our series Debunking Myths is in charge of spreading it.

Why what is proven cannot be denied without evidence

A single study, however well published it may be, is a proposal waiting for confirmation. The 2015 repetition shows that, at least in psychology, many do not hold up when someone does the experiment again. The word "proven" is reserved for what comes after. It means an idea made predictions that could fail and didn't, and that independent evidence, obtained with different methods by people who do not know each other, points to the same place. It is a practical certainty, different from that of a theorem, and enough to perform open-heart surgery or put a satellite into orbit. The shape of the Earth is confirmed by the always-round shadow it casts on the Moon during eclipses, the different stars seen from each hemisphere, the routes of ships and planes, and photographs taken from space. To deny it, one must assume that all those pieces of evidence fail at once and in the same direction.

That exam has already been taken thousands of times with the shape of the Earth and with gravity, which any classroom can measure with a pendulum. The evolution of living beings, the microbial origin of infections, the link between smoking and lung cancer, the innocence of vaccines in autism, and global warming due to human activity—which the United Nations scientific panel on climate deemed unequivocal in 2021—all passed it. None of those results are up for a vote.

The usual objection is that science changes its mind. Isaac Asimov dedicated an essay to this, The Relativity of Wrong. Believing the Earth is flat is an error, he reasoned, and believing it a perfect sphere is also one, because it is flattened at the poles and its diameter is 0.3% larger at the equator. But anyone who thinks the two errors are of the same magnitude is more wrong than both combined. Where evidence is abundant and comes from different fronts, science advances by refining decimals. Einstein corrected Newton on Mercury's orbit, and bridges are still calculated with Newton because he is correct there. Where independent evidence is scarce, no matter how much consensus there is, there is still room for reversals like that of the stomach ulcer, which is recounted below.

Denying such a result is permitted. The price is bringing better evidence than what is already on the table. Gathering it requires training and years of work. Whoever manages to pay that price ends up being right and, with luck, a Nobel prize winner.

"Well, that's my opinion" is a useful phrase for talking about a movie or a recipe. Measured facts remain outside the realm of opinion. An opinion that contradicts them is an error by another name, no matter how many followers the person expressing it may have. An Instagram reel only needs ten minutes of recording. A rebuttal requires data.

How the types of science differ

Sciences are classified using several criteria at once, and from that come labels that often get mixed up. By what they study, they are formal or factual. Formal sciences, such as mathematics and logic, work with ideas and are proven through demonstration, without going through the circuit above. Factual sciences work with facts and are divided into natural, from physics to biology, and social, from economics to psychology.

By the way they work, we distinguish between theoretical science, which builds models and derives predictions from them, and experimental science, which provokes the phenomenon under controlled conditions to see if the prediction is fulfilled. When the phenomenon cannot be provoked, one must observe with method, as astronomy and epidemiology do.

The last criterion is utility. Basic science seeks to understand, without an application in sight, and applied science seeks to solve a specific problem. Medicine and technology then turn that knowledge into treatments and machines. Food science takes it to the kitchen and computer science to every screen, with one foot in the formal sciences, because its foundations are mathematics and logic. The boundary between basic and applied science is blurry. More than one idea that seemed useless ended up, decades later, inside a phone.

Self-produced. The division between formal and factual sciences comes from Bunge (1977). In the table, factual sciences appear as natural and social. The distinction between basic and applied science comes from the Frascati Manual of the Organisation for Economic Co-operation and Development (OECD, 2015). The last four rows are fields of applied science.

The history of the ozone layer passes through several of these boxes. In 1974, the Mexican Mario Molina and the American Sherwood Rowland deduced on paper that chlorofluorocarbons (CFCs), gases used in aerosols and refrigerators, could destroy ozone in the stratosphere. It was theoretical and basic science, done with equations. The observation came in 1985, when three British researchers published that ozone over Antarctica fell sharply every southern spring. Two years later, the Montreal Protocol was signed to phase out those gases. That chemistry had moved from paper to an international treaty. The 2022 scientific assessment estimates that, if current policies are maintained, the ozone layer will recover its 1980 values around 2040 for most of the planet and around 2066 over Antarctica.

What science is not

A good part of the noise around science arises from four very repeated misunderstandings.

It is not a belief

A belief can be held without evidence and survive evidence to the contrary. A scientific claim lives as long as it overcomes attempts to refute it. When it fails repeatedly and a better explanation appears, it falls, although sometimes it may take years. That is why the phrase "I believe in science" sounds good but says little. Whoever utters it trusts, for good reason, in a system where everyone can look for the mistakes of others and in which errors, sooner or later, are put in writing.

It is the same trust you place in the airplane you did not design and the anesthesiologist you did not examine. It differs from obedience in that the evidence remains visible to anyone with the desire and means to review it.

It is not a choice

You can choose a soccer team or a political party. A virus does not ask what your opinion on vaccines is before infecting you, and gravity acts the same on those who study it as it does on those who deny it. Choosing not to believe in a measured fact changes your opinion. The fact remains where it was.

The choice comes later, when it is time to decide what to do with what has been discovered. Science estimates what effect each decision will have, and making that decision is up to ethics and politics.

It is not a finished book

All ordinary matter, that of the stars and that of your own body, makes up around 5% of the matter and energy of the universe, according to the standard cosmological model. The other 95% is divided between dark matter and dark energy, two names given to what is not yet understood. There is still so much to discover, and scientists are the first to say so.

That open frontier coexists with a solid core. A map with blank areas still preserves the coastlines already drawn. The Earth will continue to orbit the Sun on the day it is known what dark energy is, because what remains to be discovered is added to what is already known and refines it.

It is not an enemy of religion

The idea of an eternal war between science and faith was popularized in the 19th century with two books, one by John William Draper in 1874 and the other by Andrew Dickson White in 1896. Historians of science have been dismantling it for decades. Clashes occurred, and the most remembered is the condemnation of Galileo by the Inquisition in 1633. These are episodes within a much more complex history.

Christian scholars of the Middle Ages taught that the Earth was a sphere. Gregor Mendel, an Augustinian friar, discovered the laws of inheritance by cross-breeding plants in his abbey's garden. The theory we call the Big Bang today was proposed by a Catholic priest, Georges Lemaître. According to his biographers, he took steps in Rome to stop the Pope from presenting it as support for creation, which Pius XII had done in a 1951 speech. And Pakistani physicist Abdus Salam cited the Quran in his 1979 Nobel banquet speech.

A very widespread position maintains that science can only deal with what can be measured and that the meaning of life or the existence of God fall outside the reach of its instruments. Paleontologist Stephen Jay Gould summarized it by speaking of two non-overlapping magisteria. John Paul II wrote in 1996 that the theory of evolution is "more than a hypothesis" and recalled that one truth cannot contradict another.

A survey of 9,422 biologists and physicists from eight regions of the world concluded that, in most of them, scientists are less religious than their neighbors. Even so, more than half declared themselves religious in India, Italy, Taiwan, and Turkey. Those who saw science and religion in conflict did not exceed 35% in any region. In the United States, 51% of surveyed members of the American Association for the Advancement of Science (AAAS) declared in 2009 that they believe in God or a higher power, compared to 95% of the general population. For the majority of scientists surveyed, choosing between science and religion is a false dilemma.

The clash appears when a sacred text is asked to function as a geology manual. If a literal reading maintains that the Earth is six thousand years old, that is already a testable claim about the world, and measurements say otherwise. There are scientists, such as biologist Jerry Coyne, convinced that the incompatibility goes further. That debate is philosophical and remains open. While it is being discussed, many people pray in the morning and sequence genomes in the afternoon.

Four examples of science in motion

Science changes and living beings do too. Newton and a stomach ulcer speak to the former. A jar of bacteria and a fish with a wrist speak to the latter.

From Newton to the clock on your phone

For more than two centuries, Newton's gravity predicted eclipses and allowed Neptune to be found with pencil and paper before seeing it through a telescope. It failed in the orbit of Mercury. Astronomer Urbain Le Verrier verified in 1859 that its closest point to the Sun advanced slightly faster than the equations allowed, a discrepancy later calculated at 43 arcseconds per century. In 1915, Albert Einstein applied his new theory of gravity, general relativity, to that problem, and the calculation gave 43. Newton was turned into a special case of a broader theory, valid for almost everything that happens on a human scale.

Einstein's equations work for you today every time you open a map on your phone. The atomic clocks of Global Positioning System (GPS) satellites lose about 7 microseconds a day due to the speed at which they move and gain about 45 because gravity is weaker at that altitude. The net result is 38 microseconds per day. Without correcting it, the position error would grow by more than 11 km each day. That is why clocks leave the factory already adjusted.

The ulcer that was cured with antibiotics

Until the eighties, medicine attributed stomach ulcers to stress and lifestyle, and treated them by reducing acid. The drugs would close them and the ulcer would return. Two Australian doctors, Barry Marshall and Robin Warren, published in 1984 that they had found an unknown bacterium in the stomachs of 58 out of 100 patients, including almost all those who had ulcers. Few colleagues accepted that anything could live in such acid. Marshall drank a culture of the bacterium. Ten days later, he had gastritis confirmed by biopsy.

Medicine took a decade to surrender. Trials were needed that showed that, once the bacterium was eliminated, the ulcer stopped coming back. In 1994, a panel from the U.S. National Institutes of Health recommended adding antibiotics to the treatment of ulcers associated with the bacterium, which had been christened Helicobacter pylori. The Nobel Prize arrived in 2005. Acid remained in the explanation and in the treatment. The underlying cause changed, and it changed through the regulatory path, with a test that other doctors were able to repeat on their own patients.

Evolution observed in twelve jars

In 1988, biologist Richard Lenski distributed the same strain of Escherichia coli bacteria into twelve jars and began transferring a drop from each culture into a new jar every day. The experiment is still ongoing. It has exceeded 80,000 generations, which in humans, at 25 years per generation, would be equivalent to about 2 million years. Every 500 generations, a sample is frozen, so that the descendants can compete against their own ancestors.

Around generation 31,500, one of the twelve populations began to feed on citrate, a substance that had been in the jar from the first day and that this species cannot use in the presence of oxygen. By thawing old samples and repeating history, the team verified that the novelty depended on at least one prior mutation, which emerged before generation 20,000. Here, evolution could be rewound and seen again.

The fish that theory announced before it appeared

If four-legged animals descend from fish, an intermediate form must have existed. With the fossils known at that time, it had to be in freshwater rocks about 375 million years old. The team of paleontologists Neil Shubin and Ted Daeschler searched for where such rocks emerged and chose Ellesmere Island in the Canadian Arctic. There, in 2004, Tiktaalik roseae appeared, a fish with scales and fins, but with a mobile neck and a functional wrist.

In 2010, footprints attributed to four-legged animals were described in Poland, some 10 million years older than the first known fish of the Tiktaalik group. Some specialists doubt that reading. If it holds, the transition began earlier than previously thought and Tiktaalik would be a late relative of the first walkers that retained the intermediate design. The anatomy that the theory announced appeared where it was looked for. The calendar remained under revision.

In defense of science and those who do it

A researcher spends four years measuring an effect that perhaps does not exist. When she finally has a result, she hands it over to colleagues whose job is to look for errors in it. The world dedicates the equivalent of almost nine million full-time people to that trade, according to the count by the United Nations Educational, Scientific and Cultural Organization (UNESCO) with 2018 data. Most will never make a headline. We make room for some of those lives in our section Heroes of science.

Working alongside them is the person who explains that science to thirty teenagers on a Monday first thing in the morning and the person who defends it in an office against a myth that arrived via message five minutes earlier.

That work can be counted in lives. Of the deaths that vaccines have prevented in half a century, an estimated 146 million were children who had not reached their fifth birthday. One can also look up, where the ozone hole is closing because the world listened in time to two chemists and those measuring the Antarctic sky.

Those who despise science usually do so with a phone in their hand and a vaccination certificate in the drawer. They can afford to because others did the work.

A portion of those who attack science get paid to do it. The business is old. In 1969, an internal document from the tobacco company Brown & Williamson left written that its product was doubt, because sowing it was the best way to compete with the facts that the public already knew about tobacco. The doctor who sounded the alarm about vaccines and autism in 1998 had planned businesses to exploit it, including diagnostic tests with which he expected to earn up to 28 million pounds a year, as later documented by the BMJ journal.

Today, that business fits into a phone. A 2024 study analyzed three wellness influencers who frightened their Instagram followers with vaccines and then sold them health products and services. Another, on 59 groups dedicated to spreading falsehoods about vaccination, showed that their activity depends on the tools with which the platforms themselves allow money to be made. The lie that outrages is shared more than the truth that bores, and in a business that lives on attention, that is money.

The rest deny for free what they do not understand. Not understanding how a vaccine or a satellite works is normal and can be fixed. The problem begins when that lack of understanding is taken as proof. Five studies published in 2022 compared what people knew about seven issues with broad scientific consensus with what they believed they knew. Those who most opposed the consensus were those who knew the least and were the most convinced that they knew.

The most serious part comes when that confidence without study makes decisions for others. The anti-vaccine movement is the clearest case. The World Health Organization (WHO) included vaccine hesitancy among the ten threats to global health in 2019. Whoever rejects the measles vaccine for their child also decides for the baby next door, who is not yet old enough to receive it, and for the cancer patient whose chemotherapy has left them defenseless. Thus, a private error becomes an outbreak and an outbreak becomes the loss of what a country took decades to achieve.

Doubt is welcome, because it is the starting point of the method. The flat-earther in the video is right not to accept something just because an authority says it. The other half is going out to measure and accepting what comes out.

In the same laboratory, people work who would not agree on almost anything else. They understand each other because a measurement says the same thing in Lima as it does in Seoul, for the believer and for the atheist. Here, it is routine for a stranger on the other side of the world to correct you and, in doing so, do you a favor.

Faced with any claim, including those on this page, critical thinking asks that you ask yourself how it is known and what evidence would make you change your mind. If no imaginable evidence can make you change, what you have is a belief and it is worth knowing it.

The method that is discussed today in a one-minute video is the same one that emptied the polio wards and put that phone in the hand of the one recording it. Eratosthenes measured a shadow more than 2,200 years ago, and any teenager can verify today if he was right. That inheritance belongs to everyone, and its only rule of admission is to show the evidence.

What fact ever forced you to change your mind? Tell us about it at @science.driven. The best answers are turned into articles, and we say who they came from. If you want to hear about what comes next, leave your email at the bottom of this page.

What science is in an infographic