There is a doctrine of "Just War".
A Just War is one that is justified by reference to ethical principles. As a sometime egoist--though not, I hasten to add, an objectivist of any variety--I am faced with the problem of justifying war or condemning it. No one I respect, including people like Sigfried Sassoon and Eric Blair (George Orwell), condemns all war. Most thinking people recognize that there are times when we are justified in doing something called war. Given this, I'm going to take up the task of looking for justification on the assumption that one exists.
Before asking about justification, it seems reasonable to ask, "To what in reality does WAR refer?"
Reflecting on the instances I'm aware of (for example: Xerxes war against the Greeks, the Peloponesian War, Alexander's campaigns, various Roman wars, Belisarius' campaigns, Charlemange's campains, the Wars of the Roses, the English Civil War, the American Rebellion, the Crimean War, WWI and WWII, the Gulf War, etc.) I notice that all of them involved a good deal of violence. This violence has two characteristics: it was _organized_ and it was _indiscriminantly aimed at a group of people_. By "indiscriminant" I mean two things: that all one has to be is a member of one group to be considered an enemy by the members of the other group, and that the violence itself is of a widely destructive character, generally focussed on killing members of the other group but not limited to that. In war, the death of innocents has rarely been considered a matter of concern.
Even in that most limited of wars between York and Lancaster for possession of the English crown (and the head upon which it rested) it was pretty much open season on the enemy. But a war is not just distinguished by indiscriminant acts of violence between a particular groups. It is also distinguished from a feud, say, by the high level of organization involved.
Ignorant armies may clash by night, but they put a lot of effort into keeping their appointments--prior to the battle of Issus, for example, Alexander and Darius marched their armies in circles for days looking for each other, like Calvin and Hobbes going to war: "First, we have to find the enemy army..." More than one observer has commented that if the degree of cooperative effort required for two armies (or navies) to join battle had been invested instead in peace, the world might be a better place.
Violence in the sense I'm using it is a type of action, and wherever there is action it is worth considering what is acting, for this too is part of the definition of war. War is indiscriminant violence by two or more groups of individuals against each other (it does take two to make a war--when only one group is engaged in indiscriminant violence against another the concept you need to define is SLAUGHTER or MASSACRE.)
It is significant that war is the action of a group: war is a collective phenomenon. If my neighbor and I get into a fight, we are not making war on each other. Even if we attack each other with guns, we are not making war on each other. Only when we get two gangs together and give ourselves leave to use indisciminant violence against _the other gang_, considered as a whole, are we making war.
Police-work is not war because the violence is selective and limited. Gang warfare is, plausibly, war. I do not believe there is anything gained in the definition of WAR by adding that the groups involved must call themselves governments.
There are many sub-types of war: cold, limited, geurilla, and so on. All of these have some of the charateristics of war, but push toward one extreme or the other of the dimenions along which it is defined. Cold wars are fought by proxies, limited wars are less indescriminant than their plausible alternatives, geurilla wars require less organization than conventional ones.
So in my quest for a justification for war, my definition of WAR will be:
WAR: A social condition in which the members of two or more groups accept that the use of indiscriminant violence against (some of) the other group(s) is the norm, and organize themselves for the purpose of bringing this violence about.
The genus is "social condition", which makes war's collective nature explicit. The differentia is the rest. Note that despite war's social nature, I have referred to individuals ("members of") explicitly--war is not some mystical "health of the state" but a perfectly ordinary social condition.
What could justify such a condition?
The first and foremost answer is "self-defense." If I am being attacked because of my membership in some group, I am justified in banding together with other members of that group and we are then justified in organizing ourselves to do whatever is necessary to stop our attackers from attacking us. In the absense of safe, effective, alternatives, killing them is one way of doing this.
To really make this justification stick, the threat of repeated or continued attack must be high, and the alternative responses must be unlikely to have any effect. By the time Britain and Canada entered WWII, for instance, it was pretty clear that the Germans were not going to respond to anything but violence. By the time the U.S. got around to doing the same thing two years later, it was _really_ clear that Germans were unlikely to respond to anything but total victory, and that became the Allies' aim.
The indiscriminant nature of war, however, makes it hard to know if we are killing the right people. Perhaps the person with the rifle in the trench across the way is really a secret sympathizer to our cause, just waiting for a chance to cross over to our side. Are we justifed in killing him (for some reason it has generally been a him) even though he might not personally be our enemy?
This is where things come to a point, for an egoist. War is a collective phenomena, much like "the market" is. Are we justified in asking questions about individuals in these circumstances? Do we need to know if the person we are trading with in the market is a good person who will use the money we give her well? Is this a remotely valid analogy?
The thing is, while my definition mentions the members of one group explicitly, it only mentions the other group(s) qua group, not as individuals. This may be the source of the problem.
At this point, I'm stuck, and will leave my investigation for others to consider. I'll write more about it when I can see more deeply into it.
I'd like to point out, however, an important difference between what I'm doing here and what I've seen recently in various forums. I am not at all sure what answer I am going to come up with to this enquiry. My biases should be clear: I think war is sometimes justified, and I think it is sometimes justified to kill people who are not members of the enemy armed forces in the prosecution of a just war. I think it is sad but acceptable that the actions of the enemy sometimes pose such a threat to us that we are justified even in doing things that will certainly kill people who have no role in making war what-so-ever, because to spare those people would be to risk losing the war, and if there is any justification for fighting surely there is a justification for fighting to win.
But those are my biases, my prejudices, not my conclusions. I draw no lines in the sand, saying, "This must be true." To do so would be to end philosophical inquiry. And because I am doing philosophy, I'm trying to keep this in general terms. It is perfectly possible that I will conclude in the end that war can be justified, that non-combatant casualties can be justified, and that the United States and its allies would be wrong to take any military action that might harm a single hair on an Afgan grandmother's head.
I doubt it, though.