as gregory bateson commented in one essay in the seminal work "Towards an Ecology of Mind" a lot of science is deployed in a very ad hoc way - so one example of this is the treatment of scizophrenics by use of electric shock therapy - it was noticed that
monkeys that were schizoid and epileptic had a reduction in symptoms of psychosis after a fit. it was noticed that shocks induced fits - so some bright spark (pun intended) had the idea of shocking human psychotics in the vain attempt to cure them. the rest is a sad sad history
there are other examples where the arbitrary relationship between symptoms is used, metaphorically, to establish a bogus cause/effect, which, at first, has the appearance of effectiveness...
they should all be fired.