A curious defence of the view that we have a non-consequentialist moral status – that is, a status the having of which explains why it is morally wrong to treat persons in certain ways irrespective of the consequences of doing so – appeals to the claim that it would be good if we had this status, and infers that therefore, we indeed have such a status. Thomas Nagel, who gives what is probably the most well-known version, summarizes this defence of the claim that we have a non-consequentialist moral status as follows: ‘[t]he argument is that we would all be worse off if there were no rights – even if we suffered the transgressions which in that case would not count as violations of our rights – ergo, there are rights’.Footnote ¹ Nagel stresses that he is talking about moral not legal rights and that he does not take us to be in any sense capable of making it true that we have moral rights, by, for example, creating institutions protecting them, believing that we have them, or acting as if we believe we have such rights. As he puts it, the argument ‘is not supposed to be merely an argument for creating or instituting rights, through laws or conventions’; rather, ‘the argument is supposed to show that the morality which includes rights is already true’.Footnote ²

While admitting that this is ‘a curious type of argument’ which is not ‘in general . . . cogent’, Nagel nonetheless suggests that ‘it may have a place in ethical theory, where its conclusion is not factual but moral’.Footnote ³ Nagel himself does not offer further support for this contention, and critics have been quick to dismiss it as mere wishful thinking.Footnote ⁴ In a recent discussion of this argumentative strategy, David Enoch makes two suggestions.Footnote ⁵ First, ‘the question whether a given argument of [the form “wouldn't it be good if p, therefore, p”] is valid is best reduced to another question, one about the logical nature of the operators involved, and in particular whether they satisfy the analogue of the (suggested) modal-logic axiom ◊◊p → ◊p’; second, Enoch suggests more tentatively that it is plausible – or at least, not implausible – that an ‘it is good that’-operator, abbreviated by ‘G’, in fact satisfies the suggested axiom, that is, that GGp entails Gp.Footnote ⁶ The resulting proposal is that while inferences of the form ‘Gp, therefore, p’ are indeed not valid across the board, they are (perhaps) valid whenever p is equivalent to Gq, because then the inference is an instance of GGp → Gp.

In this note, I focus on Enoch's second suggestion, viz., that GGp entails Gp. It is worth emphasizing that G is a non-factive operator, given that Gp does not entail p. That is to say, we should understand ‘Gp’ as standing not simply for ‘it is good that p’ but rather as standing for ‘it is, or would be, good that p’, leaving open whether or not p is true.Footnote ⁷ Thus understood, I do not think that it is plausible that GGp entails Gp. Here is an argument that, if successful, shows that this suggestion should be rejected.

It seems undeniable that it would be good if the actual world were better than it is.Footnote ⁸ (If this were not so, it could be good if the actual world was less good than it is, and that seems a pretty crazy view to hold.) Furthermore, if the actual world were better than it is, it would contain more good states of affairs. This is just what it is for the actual world to be a better world. Even if the value of a complete world is not a simple additive function from the value of the states of affairs that it contains, it nevertheless seems plausible that a better world contains more valuable states of affairs than any world that is less good. Therefore, it would be good if the actual world contained more good states of affairs. Furthermore, the higher the number of good states of affairs that the actual world contains, the better this world is, and this seems to apply right up to the point where every state of affairs it contains is good. Therefore, for every state of affairs that the actual world contains, it is true that it would be good if that state of affairs were good. Put differently, the truth of a proposition at the actual world guarantees that it would be good if the state of affairs this proposition represents were good. If we assume, for reductio, that GGp → Gp, it follows that every state of affairs the actual world contains is in fact good. But that conclusion is absurd; it amounts to the Panglossian view that the actual world contains only good states of affairs, and is the best of all possible worlds.

Here is a different way of approaching the problem that the GGp → Gp rule gives rise to. On the face of it, there is much in the world that could be better, and surely, it would be good if this were not so.Footnote ⁹ But it seems absurd to take this as a ground for revising what you took to be the criteria for goodness, so that as a result, you no longer believe that there is anything that stands in need of improvement. If anything, you should take this observation as a ground for improving the world in accordance with your criteria for goodness insofar as you can, and perhaps as a ground for lamenting or regretting the badness of the states of affairs that you cannot improve. Yet if GGp entails Gp, the first response can be as appropriate as the second. What needs to go, I submit, is the GGp → Gp rule.

The above argument can be challenged in various ways. One obvious target is the assumption that a better world contains more good states of affairs than a less good world. However, the argument can be reformulated without this assumption. Take the whole actual world as the relevant state of affairs represented by some proposition; since it would be good if the actual world as a whole were good, we can infer that it is good, provided GGp entails Gp. Since it would also be better for the actual world to be better, right up to the point that it is the best possible world, we can again infer that the actual world is the best of all possible worlds.

Second, the argument requires that we imagine, for some state of affairs, that this state is good in one possible world, but not in another; this, it can be objected, does not make sense. The criteria for goodness are necessary, which is to say that if a state of affairs is good in one world, then that state is good in all worlds in which it obtains. If a given state of affairs appears to differ in value across worlds, we are either mistaken about its difference in value or about its being the same state across the relevant worlds. However, if the idea of states of affairs differing in value across possible worlds does not make sense in the case at hand, then we also cannot make sense of any other argument that relies on the GGp → Gp rule – including Enoch's formulation of Nagel's argument. That is to say, either this objection succeeds in undermining both the argument I offered and the reasoning my argument was designed to criticize, or it succeeds in undermining neither.

Finally, it can be objected that although the argument I offered shows that it is not the case that inferences of the form ‘GGp, therefore, Gp’ are valid across the board, they are nonetheless valid for some values of p. Taking a cue from Enoch's original proposal, one might think for instance that whenever p is equivalent to Gq, GGp does entail Gp, because GGGq entails GGq. However, even if the latter entailment holds, it is at best unclear how we can put this result to work. We were interested in establishing first-order claims about goodness, and while the original proposal was concerned with such claims, the current suggestion is not – it leaves open entirely how we are to move from second- to first-order claims. If the GGp → Gp rule is valid for some values of p, then for this to be an interesting result, the values that p can take must include non-evaluative propositions (that is, there must be admissible values for p where p is not equivalent to Gq).

Nothing I have said here precludes the possibility that there is some restricted set of values for p that salvages the GGp → Gp rule (by invalidating the argument I have offered) without making it uninteresting. But what are the criteria for inclusion in this set supposed to be? Given that retaining the rule's interest requires that we can infer at least some first-order claims about goodness, making a move analogous to the one Enoch originally proposed will not do the job. As there are no other obvious candidates, this by itself warrants some amount of scepticism about whether such a restriction is in the offing. Until a proposal is on the table, we are, then, well advised to remain suspicious of the ‘curious type of argument’ we started out with.Footnote ¹⁰