Philosophy and ways of knowing · Lesson 3 of 3

Falsifiability, paradigms, and how fields change

A claim earns its standing by what it rules out. Ask what would count against yours.

8 min read · Free — no account needed to read

Falsifiability is the idea that a claim is scientific only if some possible observation would count against it. It is a demarcation rule and a practical discipline: if nothing could contradict your theory, it explains everything and predicts nothing.

It is also more complicated than the slogan. Theories are rarely tested alone — an experiment tests the theory plus the instruments, plus the assumptions, plus the conditions. A failed prediction can always be blamed on one of the others, which is why single disconfirmations rarely overturn a theory and why judgement never fully leaves the process.

Paradigm shifts: fields mostly do ordinary, cumulative work inside a shared framework; anomalies accumulate; eventually the framework is replaced rather than corrected. This is why a literature can look settled for a decade and then reorganise, and why the strongest career move — finding an anomaly — is also the least rewarded in the short run.

Replication is how a field checks itself, and several fields discovered in the 2010s that a large share of well-known findings did not replicate. That episode is the single best argument for pre-registration, open data and reporting everything you measured — module 17.

The practical upshot for you: state what would count against your claim, expect a single result to be provisional, and treat a literature as a moving argument rather than a settled record.

See it happen

Your theory explains every case in your data, including the ones that looked contradictory.

The read that arrives on its own

The theory is powerful — it accounts for everything you have seen.

The same situation, with this idea

You ask what it forbids. If every possible observation is compatible with it, then no observation supports it either, because support requires that something could have failed to happen. A theory that can absorb any result is not strong; it is unfalsifiable. The fix is to derive one prediction specific enough to be wrong, and go and look.

After this, you can

ask what your explanation forbids, not only what it accommodates.

The takeaway

A claim earns its standing by what it rules out. Ask what would count against yours.

Easy to take the wrong way

That an unfalsifiable idea is worthless. Many valuable frameworks orient inquiry without being testable — they just should not be presented as findings.

Try this

Write one observation that would count against your central claim. If you cannot, you have found the work.