In order to preserve the possibility of ethical disagreement, it seems to me that we must distinguish between two concepts which might go under the word “good” – the first-order and second-order concepts of good.

The first order concept of good is the concept we use when making moral judgements in our everyday lives. Unless one is perhaps a moral philosopher or a fundamentalist theist, most people lack a clearly articulated theory about what it is that makes something good or bad. Nonetheless, it isn’t the case that such people take good to be merely the collection of things (character traits, actions, states of affairs) that they take to be good. That would seem both circular and psychologically difficult to maintain, given that we crave ‘reasons’ for our ethical judgements. Instead, we have a complex, disorderly and often inconsistent set of theories, rules, intuitions, convictions and instincts regarding what makes something good – we might call this a largely implicit normative ‘theory’.

But if this were the only concept of 'good', it seems that moral disagreement would not be possible, since apparently conflicting propositions regarding what it ‘good’ would in fact be employing different concepts. But genuine moral disagreement does seem possible. In such situations, I'd suggest, we fall back on a second order concept of good. But what is this concept? Is it the concept which contains its first order ‘specifications’? That is, the concept of the set of first-order conceptions of goodness? This would make sort of sense of why we might care about it - we do so because we care about those first order conceptions, and wish ours to prevail. But this is fails to respect disagreement since everything both parties to the ‘dispute’ approve of becomes ‘good’ (insofar as both parties have a first order conceptions of good)!

It is tempting to think of it as something like ‘that which should earn our highest approval’ (and conversely for ‘bad’). This definition narrowly skirts circularity, but at the risk of making unjustifiably exclusionary substantive claims (a sort of response-dependent ethics, or subjectivism) – thereby also failing to respect moral disagreement.

Perhaps, then, it is primitive, as Moore thought. It is the bedrock concept in the domain of ethics which is unanalysable but which we cannot do without; to everyone competent with the discourse its meaning is immediately obvious, and its being understood is a precondition for participation within it.

Primitivising the second-order concept seems like a reasonable possibility, simply because I can't think of plausible ways to analyse it, or do without it... Thoughts or suggestions of literature etc. which address this issue (I haven't seen any discussions of it, at least framed this way)?