In scientific research, public policy and technological applications we must often choose among scientific theories.  You might have thought we can sometimes do this by ranking the alternatives on the basis of their various merits and demerits, perhaps differently according to our purpose, and selecting whichever is best.  This familiar and sensible idea faces what to some has seemed a serious theoretical limitation.  Casting choice criteria in the role of voters, who rank the alternatives from better to worse, Samir Okasha reinterprets Arrow’s ‘impossibility’ theorem of social choice as an argument that there are no acceptable theory choice rules that maximize overall theoretical merit.

In fact, Arrow's theorem does not threaten theory choice - even when, as will often be the case, ordinal information is all there is to go by.  This is because the analogue in theory choice of one of its basic assumptions is quite unsuitable.  Indeed, well-known positive results enable us to see that, on the contrary, simple and intuitive ranking procedures often will do, such as counting one theory as better than another if it is better by most criteria.  The theory of social choice does not limit the possibilities for maximizing overall theoretical merit.  On the contrary, it tells us where some of these possibilities lie.