When Griffin wants to show the difference between the concept of well-being and other senses in which a life can be said to become good or go well, he often talks about the peculiar type of independence that obtains between well-being evaluations and moral evaluations.
This is note 38 at p. 38 of ch. II. In this note, Griffin is explaining why he can endorse a restricted version of the interest theory of value. What Griffin has troubles showing, here and in other passages dedicated to the distinction between the rational and the moral life, and prudential value, is that there is "a conception of a valuable life independent of, not subordinate to, right".
“I have followed the utilitarian assumption – actually is far more widespread – that we have a notion of good or valuable independent of our notion of what it is right or wrong to do. […] But there is a tradition that holds that good has status only in the context of a theory of right, or that the concept of right is prior to the concept of good. […] But the desire account qualifies the independence of good and right in certain ways. One thing that people with mature values take as an aim is acting morally. Ethical push comes to incorporate much of ethical pull. Moreover, when a person faces a conflict between his own interest and his moral obligation, the latter wins: moral reasons trump prudential reasons. But there still remains a conception of a valuable life independent of, not subordinate to, right.[What is the argument for this claim? Does the next sentence explain or justify it?] That immoral desires make no claim on fulfilment, that they have no authority in determining action, which seems to be the nerve of Rawls claim, does not mean that they cannot make one better off. [Apparently the synonimity between "being better off” and having well-being or a prudentially valuable life is assumed.
If this is right, the sentence with which Griffin distances himself from Rawls' position can be translated into the following claim: immoral desires have no authority in determining action, but they are relevant in terms of well-being. So Rawls' conclusion is resisted by arguing that even if the good - conceived in action guiding terms - is subordinate to right - this yet does not show that a different concept of value, which is not value in the action guiding sense, value as it is expressed in well-being evaluations, cannot exist when it opposes morality. Against this: if well-being is defined as prudential value, then, how is it possible for it not to be action guiding? After all the concept of prudence is related to action. It must be shown that some values can be acting guiding in a different sense from the sense in which morality is action guiding. But Griffin does not want to identify this difference with the difference between acting rationally and acting morally, since he, like McDowell, conceives the boundaries of rationality to be broad enough as to include receptivity to moral reasons (see here, esp . 20 and 21). So the difference he appeal to must be one between two recognizibly different ways of being guided by values, namely having goals of well-being, on one side, and living the most reasonable life, which includes obedience to moral reasons. Griffin holds that the objectives set by these two maxims can coincide, but do not necessarily:
[this is still endnote 38] And I doubt whether even an ideal person, whose values are perfectly developed, will find that ethical push and ethical pull entirely coincide. [This is also no argument.] But certainly with the mass of humanity they do not. [Most people's interests and subjective characteristics are such that, for them, doing what is best - morally speaking - and doing what will increase their well-being most represent two different courses of action. (So this is an appeal to principle M.)
This may seem obvious. It would be obvious if well-being were considered a function of people's actual desires. But it is not at all obvious that things should be so, if we start from Griffin's idea according to which well-being is a function of "rational" or "informed" desires, desires had in idealized conditions.] In thinking about others (say the inept sadist who can rise to no richer life), we might well decide that their welfare is greater for their acting wrongly. [What ground this assertion? It is not plain common sense, in so far as we are talking about welfare, and welfare is not a folk term. Maybe Griffin is claiming that when we think about the sadist fulfilling some of his desires, we may agree that the fulfillment of some of those desires makes the sadist life better in some sense. Clearly, there can be many senses in which a sadist life can be made better by the fulfillment of immoral desires. The sadist life can go better in experiential terms, meaning that it would contain more rather than less desirable experiential mental states. It can become better in material terms, etc... But how are we supposed to know in what sense Griffin means the life of the sadist life goes "better", when he opposes this sense to better in moral terms? Griffin needs to prove that there is a sense in which the sadist's life goes better by acting wrongly, which is more general than material conditions or pleasure, and less general than the concept of valuable life as it appears in this post (quote 19), that is, as a notion of choiceworthiness. ] Similarly, in thinking about our own lives, we may decide that, though we have no sufficient reason to act immorally, it would increase our well-being to do so. END OF ENDNOTE 38
For a rielaboration of those arguments, see also this post and this post.