What is falsifiability?
There is an odd but sturdy way of trusting a claim: the route to refuting it must be open. A claim that no observation could shake may look strong at first glance, but it actually teaches us nothing.
Short answer
Falsifiability is the possibility of determining in advance which observation or experimental result would refute a claim. For the philosopher Karl Popper this is the essential feature that separates a scientific claim from others: a scientific claim takes a risk, accepting that particular observations could falsify it.
The heart of the criterion: "which observation would refute me?"
At the start of the 20th century Karl Popper noticed something: some doctrines could explain every event. Whatever news arrived, the theory appeared to be confirmed, and its supporters counted this as strength. Popper argued that it was a weakness: a claim that no possible observation could refute says nothing about how the world is not, and its content is therefore empty. From this his famous criterion followed: for a claim to count as scientific, the observations that could falsify it must at least be describable.
Being falsifiable is not being false
The two ideas are often confused. Falsifiability is not a fault but a virtue: it shows that a claim is clear and bold enough to come into contact with observation. The claim "tomorrow the weather will be some way or other" cannot be falsified and is worthless; the claim "tomorrow at noon the temperature will exceed 30 degrees" can be falsified and carries information precisely for that reason. The strongest scientific theories are the ones that take the most risk: they say that countless observations could refute them, and they have nonetheless not been refuted.
A test from history: the 1919 eclipse
Einstein's theory of general relativity made a bold and risky prediction: starlight passing near the Sun should bend by a specific, calculable angle. This prediction could be tested only when stars at the Sun's edge could be seen, that is, during a total solar eclipse. Observations made at the 1919 eclipse showed that the measured deflection matched the theory. For Popper the episode was exemplary: the theory had openly entered a test that could have refuted it, and had passed. Had the measurement come out otherwise, the theory would have been seriously shaken; the risk was real. We explain how eclipses form in a separate guide.
The escape hatch: explanations added after the fact
When a claim fails a test there are two roads: revising the claim, or inventing a new explanation to rescue it. Explanations added afterwards that are themselves untestable, such as "the device was faulty that day", "the energy does not work in the presence of sceptics" or "the effect appears only in believers", are called ad hoc hypotheses. The occasional correction happens in science too; but if a doctrine closes off every failure with a new evasion, it has effectively lost its falsifiability. The flexible language of horoscopes, able to accommodate any outcome, carries the same problem; we show how that flexibility works in our guide to why horoscopes feel personal.
A tale of two planets: Neptune and Vulcan
The fate of a theory faced with the threat of falsification is nicely told by the story of two planets. In the 1800s Uranus was straying from the orbit Newton's theory predicted. There were two options: either the theory was wrong, or something had not been taken into account. Astronomers tested the second: they calculated the position of an unseen planet that would explain the deviations, and in 1846 Neptune was found in the predicted region. This was not a passing evasion, because the assumption of an "unknown planet" had itself been turned into a testable prediction. The same method was tried for the small deviation in Mercury's orbit: an inner planet named Vulcan was predicted, searched for over decades, and never found. What finally explained the deviation was general relativity, which went beyond Newton's theory. Together the two stories show that science is an art of balance, one that knows both how to defend its theory and how to move past it when it must.
Falsifiability in everyday thinking
The criterion is not just for laboratories. You can apply it to your own convictions: "if this view of mine were wrong, what would show me that?" If the question has no answer, what you hold is not knowledge but an untouchable assumption. If it has an answer, that observation must be sought honestly, and this is the best known protection against confirmation bias. It works in discussion as well: asking the other person "what would convince you?" shows from the outset whether the conversation will be productive.
Frequently asked questions
Is every unfalsifiable claim nonsense?
No. Popper's criterion draws a boundary of scientificity, not of meaning. Statements of belief and value may be unfalsifiable and still carry meaning in personal life; they simply cannot claim the status of scientific knowledge.
Is a theory thrown away the moment it is falsified?
In practice, no. First the possibilities of measurement error and experimental flaw are eliminated; if the contrary result is independently replicated, the theory is corrected or replaced by a more comprehensive one. Newtonian physics being surpassed by relativity is an example.
Is astrology falsifiable?
In certain forms, yes: claims such as predicting personality from a birth chart can be tested and have been tested, and the results have not exceeded chance level. The problem is that the language of daily readings is usually flexible enough not to conflict with any outcome.
Is falsifiability the only criterion in science?
No; the philosophy of science also discusses further criteria such as repeatability, explanatory power and fruitfulness. Falsifiability stands out because it is the most practical and the easiest to carry over into everyday thinking.
This guide draws on Karl Popper's work in the philosophy of science and on the content of sources in the history of science.