Karl Popper proposed the idea of falsification, that rather than starting with observations, scientists should begin with theories. And in the testing of theories, rather than trying to prove them true, the goal is to prove them false.

Any theory that is proved to be false can be modified and retested, or rejected. And in that way, science progresses by finding out whether theories are able to hold up, and can be retested to see whether or not they can be falsified. While there’s always the chance for results to continue to support a theory, if they are found to not hold up, then that calls them into question. This comes in very useful with cases of pseudoscience and irrelevant claims, as they’ll need to have a theory that can be tested.

Naturally there could be problems with using falsification, such as an error in testing of a theory or even gaps in our knowledge that make it difficult to test a theory. A proposal by Thomas Kuhn is that there’s a certain weight of evidence for certain theories and a whole new paradigm may emerge that requires a new way of interpreting evidence, or even new problems that require solving. 

While scientific developments can involve theory-based speculations leading to confirmation or retraction, there’s still the possibility for new developments and changes that could overhaul what we already know.