Books on Amazon
Theory/Feyerabend: there is no one single interesting theory that corresponds with all known facts in its field.
Theory/Feyerabend: any theory that contradicts another outside of the observations is supported by exactly the same observations and therefore also acceptable.
Theory/Hume/Feyerabend: Theories cannot be derived from facts. If we are to demand that only theories be admitted that follow from the facts, no theory remains at all.
Theory/Feyerabend: a theory may not be incompatible with the data, because it is not correct, but because the data are contaminated! They could contain, and analyzed, perceptions, which correspond only partially to external processes, or be clad in old conceptions, or judged by backward auxiliary sciences.
Theory/Observation Language/Feyerabend: even the most thorough examination of an observation sentence does not disturb the concepts by which it is expressed or the structure of the perception image. How can we examine something that is constantly used and presupposed in every statement?
Lakatos: pro ad-hoc theories.
Theory/Popper: new theories have - and this must be so - excess content, which - and this should not be so - is gradually infected by ad-hoc adjustments.
Theory/Lakatos: new theories are - necessarily - ad-hoc. Excess content is created bit by bit by gradually extending the theories to new facts and areas.
Wider den Methodenzwang Frankfurt 1997
Erkenntnis für freie Menschen Frankfurt 1979