Philosophers of science dealing with axiomatic (deductive) systems have
devised what may be called the magician's hat model of science. A clue is
provided by Einstein's definition of method:
"Guided by empirical data, the investigator rather develops a system of
thought which, in general, is built up logically from a small number of
fundamental assumptions, the so called axioms."
The problem is that Einstein's own "building up logically" is
incomprehensible not only to philosophers but also to Einstein himself and
anybody else. So, in philosophical construals, everything
concerning "building up logically" is taken to constitute the interior of a
magician's hat where you put ties which are then turned to rabbits. The hat
needs some camouflage of course. Neopositivists define the steps
in "building up logically" as analytical, Popper restricts deduction to the
process in which a wolf rather than a rabbit jumps out of the hat,
Feyerabend vindicates the existence of the hat in a world where "anything
goes" etc.