If science is sometimes wrong, why should we trust it? The scientific method explained

Science doesn’t seek “truth” — it seeks to prove itself wrong

If science is sometimes wrong, why should we trust it?

Because science isn’t about being right — it’s about being less wrong.

The scientific method doesn’t try to prove theories are true. It tries to prove they’re false. What survives the tests (for now) is what we use.

When better evidence appears, science changes — and that’s not weakness, it’s science’s greatest strength.

No other method of knowledge self-corrects so well.

And when I understood this, I stopped seeing “science changed its mind” as a problem — and started seeing it as a solution.

Note: This post is a deep dive into “Can science explain everything?”. If you haven’t read it yet, I recommend starting there.

---

What science really is (and isn’t)

Before explaining how it works, I need to destroy a myth:

Science isn’t “absolute truth.”

Science is a method of building models that are progressively less wrong about reality.

Examples:

Newton (1687):

• Gravity is a force that pulls objects
• Works perfectly for 99% of everyday situations

Einstein (1915):

• Gravity is curvature of space-time
• Works better than Newton in extreme situations (very high speeds, intense gravitational fields)

Was Newton wrong?

No. He was incomplete.

Einstein didn’t “overthrow” Newton — he expanded the scope. Rockets still use Newtonian physics. GPS uses Einstein’s relativity.

Both are useful models in different contexts.

---

The scientific method: how it works in practice

Here’s the core of how science actually works.

Step 1: Observing a problem

You notice something you want to explain.

Example: Why do objects fall?

Step 2: Formulating a hypothesis

You create a testable explanation.

Example: “There’s a force (gravity) that pulls objects downward.”

Step 3: Deducing consequences

If the hypothesis is true, what should happen?

Example: “If gravity exists, heavier and lighter objects should fall at the same speed (ignoring air resistance).”

Step 4: Experimental test

You test the prediction.

Example: Galileo (supposedly) dropped two spheres of different masses from the Tower of Pisa. They fell together.

Step 5: Attempting falsification

Here’s where it gets interesting.

You don’t try to prove you’re right. You try to prove you’re wrong.

If the hypothesis survives refutation attempts, it’s corroborated (provisionally accepted).

If it fails, you discard or modify the hypothesis.

---

Falsificationism (Karl Popper): the soul of the scientific method

Karl Popper (1902-1994), Austrian philosopher, revolutionized philosophy of science with a simple but powerful idea:

“A scientific theory cannot be proven — it can only be refuted.”

What does this mean?

Verificationism (old method):

• We try to prove theory is correct
• We accumulate supporting evidence
• Problem: you can always find evidence that “supports” any theory (even wrong ones)

Falsificationism (Popper):

• We try to prove the theory is wrong
• If it survives honest refutation attempts, we accept it (provisionally)
• If it fails, we discard it

Classic example: swans

Claim: “All swans are white.”

Verificationism:

• See 1,000 white swans → “confirmed!”
• See 10,000 white swans → “more confirmed!”
• See 1 million white swans → “totally confirmed!”

Problem: It takes just one black swan to disprove everything.

Falsificationism:

• Don’t try to prove all are white
• Try to find a black swan
• If you don’t find one (despite looking), provisionally accept “all swans are white”
• But there always might be a black swan out there

Science works this way. We never claim “absolute truth” — only “we haven’t refuted this yet.”

---

Scientific theory must be falsifiable

Here’s the key to understanding what separates science from pseudoscience.

A scientific theory must be potentially refutable.

This means: there must exist an experiment/observation that, if it gives a certain result, disproves the theory.

Examples:

Einstein’s theory (falsifiable):

• Predicts light bends near massive objects
• Test: Observe stars during solar eclipse
• If light didn’t bend, theory would be refuted
• Result: light bent (1919) → theory corroborated

Astrology (not falsifiable):

• Predicts “Leos like being the center of attention”
• But if a Leo doesn’t like it? “Oh, must have a shy sign as rising sign”
• Always has explanation for any result
• Can’t be refuted → not scientific

Homeopathy (not falsifiable in practice):

• When it doesn’t work: “you didn’t dilute enough”, “didn’t shake properly”, “needs personalized dosage”
• Always has an out for negative results

Science puts itself at risk. Pseudoscience never does.

---

Science self-corrects — and that’s strength, not weakness

When science “changes its mind,” many people think it’s a sign of weakness.

But it’s the opposite.

How self-correction works:

1. Peer review

• Scientist publishes result
• Other scientists try to replicate
• If they can’t, result is questioned

2. Scientific competition

• Scientists want to refute each other
• Discovering error in famous theory = fame and recognition
• Result: Theories are constantly under attack

3. Experiment replication

• Experiments must be repeatable
• If nobody can repeat it, it’s not accepted

Examples of self-correction working:

Phlogiston theory (1667-1770):

• “Substances burn because they release phlogiston”
• Refuted by Lavoisier (discovered oxygen)
• Science abandoned the theory

Luminiferous ether (19th century):

• “Light propagates through an invisible medium called ether”
• Refuted by Michelson-Morley experiment (1887)
• Einstein showed light doesn’t need a medium

Spontaneous generation:

• “Life spontaneously arises from inert matter”
• Refuted by Pasteur (1859)
• We now know life comes from life

Science abandons theories when contradicting evidence appears.

Pseudoscience? Never changes. Astrology works exactly the same as it did millennia ago.

---

Scientific “errors” aren’t failures — they’re refinements

When Newton “got it wrong,” it wasn’t failure — it was an intermediate step.

The atomic model: evolution, not error

Dalton (1803): Atom is an indivisible sphere
Thomson (1897): Atom has electrons (“plum pudding” model)
Rutherford (1911): Atom has a nucleus
Bohr (1913): Electrons in defined orbits
Quantum Mechanics (1920s): Electrons in probability clouds

Each model was useful in its time. Each was refined by the next.

Was Dalton wrong?

No. He was doing his best with available evidence.

Science doesn’t get it wrong — it improves.

---

Comparison: science vs pseudoscience

Feature	Science	Pseudoscience
Changes?	Yes, when evidence appears	Never
Falsifiable?	Yes, always	No
Accepts error?	Yes, publicly	No, always has an excuse
Peer review?	Yes, rigorous	No
Predictions?	Specific and testable	Vague and adjustable

Practical example:

Evidence-based medicine:

• Tests treatments in double-blind trials
• Publishes results (positive and negative)
• Discards what doesn’t work
• Result: Antibiotics, vaccines, surgeries that work

Homeopathy:

• When it works: “see, it works!”
• When it doesn’t: “needs right dosage”, “everyone is different”
• Never abandons practices that don’t work
• Result: Sugar water sold as medicine

---

Questions I had (and the answers)

“So science is just guessing?”
No. It’s guessing exhaustively tested. The difference between “just a hunch” and science is that science tries to prove itself wrong. Hunches never self-question.

“Why trust something that changes?”
Because changing when evidence appears is a sign of honesty, not weakness. Dogmas never change — and they’re frequently wrong.

“Do scientists lie?”
Some do (they’re human). But the system (peer review, replication, competition) eventually exposes fraud. Cases like Andrew Wakefield (faked vaccine study) were discovered and exposed by the scientific community itself.

“How do I know they won’t change their mind again?”
They might. And they should, if better evidence appears. But the more a theory is tested and survives, the more reliable it becomes. Newtonian gravity has worked for 300+ years — it’s very unlikely to be completely wrong.

---

Why we trust science: it works pragmatically

In the end, we trust science because it works.

Technology is empirical proof:

• GPS works → Einstein’s Relativity is correct (at least sufficiently)
• Vaccines work → Immunology is correct
• Planes fly → Aerodynamics is correct
• Electronics work → Quantum mechanics is correct

If quantum physics were “just theory,” your smartphone wouldn’t work.

Science doesn’t need to be “absolute truth” to be useful. It needs to be a useful model — and it is.

---

Why this fascinates me (and reassures me)

Because it shows science is humble.

Science admits it might be wrong. Always might be. And when it is, it corrects.

Compare with dogmatic systems:

• Dogma: “This is true because authority X said so”
• Science: “This seems true based on evidence, but could be refuted tomorrow”

Which is more honest?

Science doesn’t promise absolute certainty — it promises the best available approximation at the moment.

And that’s far more trustworthy than invented certainties.

---

💡 Summary in 3 points:

1. Science doesn’t seek “absolute truth” — it seeks progressively less wrong models through falsification (trying to prove itself wrong)
2. Science self-corrects (peer review, replication, competition) — pseudoscience never changes
3. We trust it because it works pragmatically (GPS, vaccines, planes, electronics prove models are useful)

---

Enjoyed understanding how science really works? This post dives deeper into concepts from “Can science explain everything?” — read it to understand the philosophical limits of science and why recognizing limits is strength, not weakness.

Leia também Can Science Explain Everything Limits Epistemology

---

References:

• POPPER, Karl. The Logic of Scientific Discovery. Routledge, 2005.
• POPPER, Karl. Conjectures and Refutations. Routledge, 2014.
• Philosophy Now: Karl Popper and Falsificationism philosophynow.org
• Stanford Encyclopedia: Karl Popper plato.stanford.edu

---

Personal note: I want to study more about Imre Lakatos and the methodology of scientific research programs — he criticizes Popper saying “naive” falsificationism doesn’t explain how science actually works (scientists don’t abandon theories on first negative test). Lakatos proposes “hard core” vs “protective belt” of hypotheses. Fascinating. That’s for another post.