Falsifiability is that attribute of an assertion which makes it at least theoretically possible to prove the assertion's truth or falsehood. If there is no observation which could, in principle, show the assertion to be wrong, then there is correspondingly no opposite observation which could show the assertion to be correct.
One corollary of this is that if you (the asserter) cannot even imagine any set of circumstances which would change your mind about your assertion, then you must realize that you would make the same assertion even if it were not true – because the fact of your assertion's being false could not in any way stop you from believing that it was true – which in turn means that your assertion cannot be trusted, and bears no correlation to reality.
Note that an "observation" can be defined as broadly as necessary in order to allow for very rarefied observations, e.g. "If God exists, then there will be a Judgment Day" could be falsified – although not immediately! – by the absence of a Judgment Day prior to the extinction of humankind. (If the extinction of humankind is taken to be Judgment Day, then the assertion is completely unfalsifiable, as the only way to prove it wrong would be for humankind to survive forever and for someone to observe this.) We might say, however, that such observations are practically non-falsifiable (unless Judgment Day happens to come along and thereby proves that God does, in fact, exist as defined by the assertion).
The methods of science generally insist that any assertion must be falsifiable in order to be taken seriously.
Frequently cited examples of non-falsifiable assertions include:
- Some entities invented to highlight the illogic of much religious doctrine:
- The existence of God is often seen as non-falsifiable because no specific testable assertions are made about the nature of His/Her/Its existence.
- Direct creation theories, in which all the scientific evidence for millions of years of evolution and billions of years of geology and cosmology were created, along with the Earth, 10,000 years ago or less, are generally non-falsifiable because of the supposition that any evidence which seems to argue against them was in fact "planted" by an omnipotent being (God) who can act in ways that not only violate the laws of nature as we understand them but are perhaps even inconsistent with any laws we can comprehend.
The idea of potential falsifiability as the criterion demarcating science from non-science was first put forth by Karl Popper in his first book, Logik der Forschung (The Logic of Scientific Discovery), in 1934. [W]
- 2007-09-27 How to Convince Me That 2 + 2 = 3 by Eliezer Yudkowsky: even simple arithmetic must be falsifiable if you are to understand it on some basis other than trust/faith, otherwise you leave yourself vulnerable to disinformation such as that portrayed in 1984.