Falsifiability

What it is

A claim is falsifiable if some possible observation could show that it is wrong. "All swans are white" is falsifiable: one black swan would sink it. "Everything happens for a reason" is not, because nothing that happens could ever count against it.

The philosopher Karl Popper argued that this is what sets scientific claims apart. A scientific theory, in his view, should stick its neck out. It earns our trust by surviving tests that could have proved it wrong.¹

Falsifiable does not mean false. It means testable: you can say in advance what would count against it.

Why it matters

A claim that can't be tested can't be checked, however convincing it sounds. And a theory that can explain any possible result is not strong. It is a warning sign, because it tells you nothing about what to expect.

Philosophers still debate whether falsifiability alone is enough to separate science from everything else. So treat it as a useful test, not a complete definition of science.

Two real tests

Starlight and the Sun. In 1919, astronomers travelled to watch a total eclipse of the Sun. During an eclipse, stars near the Sun become visible. The question was whether the Sun's gravity bends the light from those stars, and by how much. The expeditions were planned to tell three possible answers apart: no bending at all, a small amount predicted by older physics based on Newton's ideas, or about twice that amount, as Albert Einstein's new theory predicted.² Einstein's theory could easily have failed. The results were reported as favouring Einstein.²

Astrology. In a double-blind test, astrologers tried to match people's birth charts to personality profiles of those same people. Double-blind means that neither the astrologers nor the people running the test knew which chart belonged to whom. Astrologers helped design the test, and they predicted they would get at least half right. In fact, they made the correct first choice 40 times out of 116. Pure guessing would have been expected to give about 38.5.³ Astrology made a testable claim, and it didn't survive the test.

In everyday life

• Lucky charms. "This shirt helps my team win." Could anything prove that wrong? If every loss is explained away ("we'd have lost by more without it"), the belief can't be tested as stated.
• Conspiracy theories. If every piece of evidence against a theory is treated as more proof of a cover-up, the theory can't be falsified. That makes it impossible to check, not stronger.
• Health claims. "This supplement supports your wellbeing" is so vague that no result could count against it. "This supplement lowers blood pressure within a month" can be tested.

The common trap

The first trap is thinking that "falsifiable" means "proven false." It only means that it could be shown false, if it were wrong.

The second trap is admiring a theory that explains everything. If no possible result could count against an idea, it isn't telling you much.

The third trap is thinking one failed test always kills a theory. In practice, surprising results get rechecked before anyone abandons a well-supported idea. See the Iteration entry.

Some ways out:

• For any claim, ask: "What would prove this wrong?"
• Be wary when every possible result is taken as support.
• Turn vague claims into specific ones that could fail.

This links closely to Confirmation bias. Testing your ideas where they could fail, not only where you expect a "yes," is the best defence against that bias.

Try it

Write down three things you believe, such as "my lucky shirt helps my team." For each one, write down the observation that would prove it wrong.

If you can't name one, or you would explain away any result, the belief isn't falsifiable as you've stated it. Try rewriting it so that it is. For example: "My team wins more often when I wear the shirt, over the next 20 games." Then keep score.

Related concepts

Confirmation bias · Iteration · Scientific uncertainty · Actively open-minded thinking · Motivated reasoning

Sources

1. Popper, K. (2005). The logic of scientific discovery. Routledge. https://doi.org/10.4324/9780203994627

2. Dyson, F. W., Eddington, A. S., & Davidson, C. (1920). A determination of the deflection of light by the sun's gravitational field, from observations made at the total eclipse of May 29, 1919. Philosophical Transactions of the Royal Society A, 220(571–581), 291–333. https://doi.org/10.1098/rsta.1920.0009

3. Carlson, S. (1985). A double-blind test of astrology. Nature, 318(6045), 419–425. https://doi.org/10.1038/318419a0