The chief persuasive argument in favour of utilitarianism has been that the dictates of any deontological ethics will always, on some occasions, lead to the existence of misery that could, on utilitarian principles, have been prevented. Thus if the deontologist says that promises always should be kept (or even if, like Ross, he says that there is a prima fade duty to keep them) we may confront him with a situation like the following, the well-known 'desert island promise": I have promised a dying man on a desert island, from which subsequently I alone am rescued, to give his hoard of gold to the South Australian Jockey Club. On my return I give it to the Royal Adelaide Hospital, which, we may suppose, badly needs it for a new X-ray machine. Could anybody deny that I had done rightly without being open to the charge of heartlessness? (Remember that the promise was known only to me, and so my action will not in this case weaken the general confidence in the social institution of promising.) Think of the persons dying of painful tumours who could have been saved by the desert island gold!

"But", the deontologist may still object, "it is my doctrine which is the humane one. You have accused me of inhumanity because I sometimes cause avoidable misery for the sake of keeping a rule. But it is these very rules, which you regard as so cold and inhuman, which safeguard mankind from the most awful atrocities. In the interests of future generations are we to allow millions to die of starvation, or still more millions to be sent to forced labour? Is it not this very consequentialist mentality which is at the root of the vast injustices which we see in the world today?" Two replies are relevant. In the first place the man who says this sort of thing may or may not be interested in the welfare of future generations. It is perfectly possible not to have the sentiment of generalized benevolence but to be moved by a localized benevolence. When this is localized in space we get the ethics of the tribe or the race: when it is localized in time we get an ethics of the present day and generation. It may well be that atrocities carried out for the sake of a Utopian future repel some people simply because they mortgage the present for the sake of the future. Here we have a difference about ultimate ends, and in this case I cannot accuse my opponent of being either confused or superstitious, though I may accuse him of being limited in his vision. Why should not future generations matter as much as present ones? To deny it is to be temporally parochial. If it is objected that future generations will only probably exist, I reply: would not the objector take into account a probably existing present population on a strange island before using it for bomb tests?

In the second place, however, the opponent of utilitarianism may have a perfectly disinterested benevolence, save for his regard for the observance of rules as such. Future generations may in fact mean as much to him as present ones. To him the utilitarian may reply as follows. If it were known to be true, as a question of fact, that measures which caused misery and death to tens of millions today would result in saving from greater misery and from death hundreds of millions in the future, and if this were the only way in which it could be done, then it would be right to cause these necessary atrocities. The case is surely no different in principle from that of the battalion commander who sacrifices a patrol to save a company. Where the tyrants who cause atrocities for the sake of Utopia are wrong is, surely, on the plain question of fact, and on confusing probabilities with certainties. After all, one would have to be very sure that future generations would be saved still greater misery before one embarked on such a tyrannical programme. One thing we should now know about the future is that large-scale predictions are impossible. Could Jeremy Bentham or Karl Marx (to take two very different political theorists) have foreseen the atom bomb? Could they have foreseen automation? Can we foresee the technology of the next century? Where the future is so dim a man must be mad who would sacrifice the present in a big way for the sake of it. Moreover even if the future were clear to us, it is very improbable that large scale atrocities could be beneficial. We must not forget the immense side effects: the brutalization of the people who ordered the atrocities and carried them out. We can, in fact, agree with the most violent denouncer of atrocities carried out in the name of Utopia without sacrificing our act-utilitarian principles. Indeed there are the best of act-utilitarian reasons for denouncing atrocities. But it is empirical facts, and empirical facts only, which will lead the utilitarian to say this.

The future, I have remarked, is dim, largely because the potentialities of technological advance are unknown to us. This consideration both increases the attractiveness of a utilitarian ethics (because of the built-in flexibility of such an ethics) and increases the difficulty of applying such an ethics. Normally the utilitarian is able to assume that the remote effects of his actions tend rapidly to zero, like the ripples on a pond after a stone has been thrown into it. This assumption normally seems quite a plausible one. Suppose that a man is deciding whether to seduce his neighbour's wife. On utilitarian grounds it seems pretty obvious that such an act would be wrong, for the unhappiness which it is likely to cause in the short term will probably be only too obvious. The man need not consider the possibility that one of his remote descendants, if he seduces the woman, will be a great benefactor of the human race. Such a possibility is not all that improbable, considering the very likely vast number of descendants after a good many generations, but it is no more probable than the possibility that one of his remote descendants will do great harm to the human race, or that one of the descendants from a more legitimate union would benefit the human race. It seems plausible that the long-term probable benefits and costs of his alternative actions are likely to be negligible or to cancel one another out.

An obviously important case in which, if he were a utilitarian, a person would have to consider effects into the far future, perhaps millions of years, would be that of a statesman who was contemplating engaging in nuclear warfare, if there were some probability, even a small one, that this war might end in the destruction of the entire human race. (Even a war less drastic than this might have important consequences into the fairly far future, say hundreds of years.) Similar long term catastrophic consequences must be envisaged in planning flight to other planets, if there is any probability, even quite a small one, that these planets possess viruses or bacteria, to which terrestrial organisms would have no immunity.

The progress of science and technology could yield many more cases which might pose dramatic problems to the moralist. Consider the moral problems which would be set by a spectacular innovation in the field of positive eugenics,[1] or perhaps of direct tampering with the human genetic material, or of a spectacular discovery which would enable the life span of man to be prolonged indefinitely. (For example, would the realization of the last possibility imply the Tightness of universal euthanasia?) Again, suppose that it became possible to design an ultra-intelligent machine[2] (superior in intelligence to any human) which could then design a yet more intelligent machine which could ... (and so on).

Consider positive eugenics first. Suppose that it did one day turn out that by methods of positive eugenics, it became possible markedly to increase the intelligence of the whole human race, without using tyrannical or unpleasant means and without reducing the genetic diversity of the species. (There are important biological advantages in diversity.) Ought a utilitarian to approve of such a measure? Clearly something will depend on whether he is a hedonistic or an ideal utilitarian. The ideal utilitarian may have an intrinsic preference for more intelligent states of mind. However the hedonistic utilitarian might agree with the ideal one if he thought that intelligence was extrinsically valuable, for example if he thought that wars and poverty were due mainly to stupidity, and perhaps if he thought that more avenues for obtaining pleasure were open to intelligent people.

Even more interesting ethical issues arise if we imagine that biological engineering went so far as to enable the production of a higher species of man altogether. Similar issues arise also if we imagine that it becomes possible to produce an ultra-intelligent artefact which possesses consciousness. (This is not the place to enter into the deep metaphysical issues which arise out of the question of whether a conscious artefact is possible or not.) Let an entity which is either a member of the envisaged superior species or is an ultra-intelligent conscious artefact be conveniently referred to as 'a superman'. What might a utilitarian's attitude be towards possible actions which would lead to the production of a superman? It is quite possible that there should be a kind of utilitarian who valued only the happiness of his own species and was perfectly indifferent to that of higher and lower species. He might even envisage the superman with fear and hatred. Such a man's ethics would be analogous to the ethics of the tribe. Suppose alternatively that he were an ideal or quasi-ideal utilitarian, who thought that it was better to be Socrates dissatisfied than a fool satisfied. Should he similarly yield ethical precedence to the superman?