I believe that it is now pretty generally accepted by professional philosophers that ultimate ethical principles must be arbitrary. One cannot derive conclusions about what should be merely from accounts of what is the case; one cannot decide how people ought to behave merely from one's knowledge of how they do behave. To arrive at a conclusion in ethics one must have at least one ethical premiss. This premiss, if it be in turn a conclusion, must be the conclusion of an argument containing at least one ethical premiss. And so we can go back, indefinitely but not for ever. Sooner or later, we must come to at least one ethical premiss which is not deduced but baldly asserted. Here we must be a-rational; neither rational nor irrational, for here there is no room for reason even to go wrong.
(Brian Medlin, "Ultimate Principles and Ethical Egoism," Australasian Journal of Philosophy 35 [August 1957]: 111-8, at 111)
Note from KBJ: Medlin is conflating at least two things. First, he is making a (correct) point about the validity of arguments. In a valid deductive argument, the conclusion contains no more information than the conjunction of the premises. It may contain less or it may contain the same amount, but it may not contain more. Arguing from factual premises ("is" statements) to an evaluative conclusion (an "ought" statement) violates this rule. Hume's Law (no "ought" from "is") is simply a special case of the general rule. Second, Medlin is pointing out (correctly) that, in order rationally to persuade someone to accept a conclusion, you must use only premises that he or she accepts. If you use a premise that is not accepted by your interlocutor, your argument gets no grip on him or her. You will have to back up (as it were) and argue for the premise, once again using only premises that your interlocutor accepts. This process can go on indefinitely, which is why rational persuasion is so difficult and time-consuming. To persuade, one must find common ground and proceed from there. In short, Medlin is conflating (1) a (correct) point about validity, which is a formal feature of argument, with (2) a (correct) point about rational persuasion, which is a human activity.
