The Virtuous Lie
Kant held that all lying is wrong, even when the consequences of telling the truth are terrible. Most people have disagreed: sometimes lying is the right thing to do in cases in which truth telling would have bad consequences (e.g. the Nazis looking for the fleeing Jewish girl). The idea is that lying is always prima facie wrong (to use Ross’s terminology) but that in some cases net utility overrides this wrongness; not perhaps any degree of negative consequence but when the bad consequences are extreme enough. Surely it is right to lie if a whole civilization is at stake! What is not considered, however, is whether it can be right to lie even when the consequences of telling the truth are not bad. Can it ever be right to lie when the consequences of doing so are worse than the consequences of telling the truth? That sounds impossible, since consequences are the only conceivable way that the prima faciewrongness of lying can be overridden; there has to be something to counterbalance the wrongness of lying—and what could that be but the securing of good consequences? The only morally permissible lie is the beneficial lie: there can be no other valid reason for telling a lie.
But this ignores another possible reason for lying: the binding force of a promise. Suppose you have a friend, Phil, who is rather short; and suppose Phil has an enemy, Bert, who is pettily obsessed with Phil’s height in relation to his own (he is an inch taller than Phil). Bert has been trying to find out Phil’s height so that he can loudly boast of his vertical superiority to Phil, but hitherto has been unable to ascertain this information. Phil asks you to promise not to tell Bert his height because he knows it will fill Bert with unseemly glee, and you agree. Bert subsequently asks you to tell him Phil’s height. If you tell the truth Bert will be overjoyed, relishing his petty rivalry with your friend Phil; but if you lie Bert will be disappointed and grumpy. Should you lie or tell the truth? If you tell the truth you break your promise to your friend, possibly being motivated by utilitarian considerations; but if you lie, you fail to maximize the amount of happiness in the world (let’s suppose Phil knows nothing of your encounter with Bert). This would be a lie that has no good consequences defined in terms of net utility. I say it would be right to lie in these circumstances, yet utility cannot be the reason. The reason is obvious: you made a promise to Phil, and that promise imposes a moral obligation on you. Now if it turned out that keeping that promise would result in the death of innocent people even Phil would agree that the promise should be broken, but not just because Bert would be made marginally less happy than the alternative. Promises must be kept even when utility is not maximized—as when you keep a promise to meet someone for lunch even though a more attractive option has presented itself. So lying can be the right thing to do just because you have promised (for good reasons) to lie—even when utility is not maximized. The promise overrides the lack of utility maximization. So virtuous lies can occur even in cases where consequences indicate otherwise. This is one of those cases in which your moral duties all things considered favor lying but not because one of these duties is to ensure the best possible outcome in respect of consequences.
But isn’t the example abnormal? We don’t usually promise to lie, though we often promise to tell the truth. That may be true as a matter of statistical fact, but it doesn’t reflect a necessary truth. Consider a society ravaged by disease in which visible signs of the disease mar people’s physical appearance. There may be a general promise, implicit or explicit, not to remark on such signs—it just hurts people’s feeling to be reminded of their condition. This promise has binding force even if it is slightly worse for people in the long run if they don’t know what they look like—not catastrophic but real. Once the promise is in effect it creates a prima facie obligation to maintain the lie, even when the consequences of doing so may be slightly worse than insisting on the truth. Promising to lie is like promising to do anything: it creates a duty to act as promised—except in cases in which the consequences are dire enough to overrule the promise. Thus there could be a society in which this kind of virtuous lying is common and expected. There will be no general prohibition in this society against lying. We can even formulate a moral rule: it is morally wrong to tell the truth if you have promised to lie (insert the usual caveats). You ought to lie if you have promised to lie, just as you ought to do whatever you have promised. So Kant was wrong even by his own non-consequentialist standards: a pure deontologist can accept that lying is sometimes right, because the rule of promise keeping applies to lying too. Lying can be required by the rules of morality even when its consequences are less than optimal. 
 Ross gives the compelling example of carrying out a deceased person’s will: if the deceased has willed his property to John, it is wrong to allocate it to Jim on the ground that you think (correctly) that Jim will be made marginally happier than John. Promising as such carries moral weight irrespective of consequences.
 It is tempting to conclude that lying is never intrinsically wrong, i.e. wrong just by being a lie. In the case of Phil and Bert you do nothing wrong in lying; the wrongness of lying, which is indeed generally wrong, depends on the surrounding context.