At first, this argument seems completely silly and erroneous; why would anyone want to make a statement that contradicts all empirical "facts" provided? Well, Feyerabend has shown through numerous arguments (some of which are posted on this blog below) how counterinduction can be a more efficient tool for scientific discovery than the usual induction.
He starts on chapter 5 that,
As an example of such an attempt I examine the tower argument which the Aristotelians used to refute the motion of the earth. The argument involves natural interpretations - ideas so closely connected with observations that it needs a special effort to realize their existence and to determine their content.The "tower argument" asserts that since an object falling from a tower moves vertically and not diagonally, the Earth cannot be moving.
This observational "fact"-- labeled by Feyerabend as a natural interpretation, was believed to be so "obvious" that theories contradicting this idea were quickly rejected.
However, Feyerabend asserts that such observational "facts" must not be taken for granted. Analyzing Galileo's works, he argues that even this "basic" fact assumes a "naive realism with respect to motion." Because relative motion from a point of view on Earth is not separated or distinguished from absolute motion, this empirical example creates a false illusion that the Earth must be moving. The latter concept of motion became a new interpretation of the situation and counterinductively allowed us to realize the flaws associated with the original assumption.
Psychologically, admitting the fact that one did not notice the object moving in such a wide trajectory as seen in the image above can be very difficult-- a situation Feyerabend calls a paradigmatic case. Even in these cases, however, the counterinductive interpretation allowed us to notice the Trojan horse behind the natural interpretation.
Thus, we cannot stick with one "accepted" theory that encompasses these "known" facts. But we cannot reject all natural interpretations, as without them a person would be "completely disoriented" and would not be able to "start the business of science." So how can we deal with such empirical "facts" and theories?
Feyerabend argues that the only logical option left is to form many theories, with some contradicting others, so that we can counterinductively analyze and determine the validity of these "facts." As he says, we need to use an "external measure of comparison, including new ways of relating concepts and percepts." This strange approach will reveal how such natural interpretations actually work.
In the example above, the Copernican view of the motion of the Earth acts as the "external measuring rod." It allows us to uphold apparent contradictions until we completely examine them, "or else the examination, the attempt to discover the antediluvian components of our knowledge, cannot even start."
Since a particular natural interpretation cannot be tested directly (by comparing them with former "results of observation"), the use of other interpretations is the only option left. Therefore, Feyerabend declares that,
Ideological ingredients of our knowledge and, more especially, of our observations are discovered with the help of theories which are refuted by them. They are discovered counterinductively.