2005

NATO morality and the Kosovo-War
An ethical commentary - ex post

Georg Meggle

0. All it took for the 37000 or so NATO air raids against Serbia/Yugoslavia in 1999 to receive such broad public blessing in most NATO-countries was the goal (genuine or alleged) expressed in the term used to describe the entire operation: it wasn't war, it was a "Humanitarian Intervention". The only fly in the ointment was the lack of a mandate from the UN. Yet this snag was actually relished by the strongest proponents of intervention, who interpreted NATO's action as a victory for morality over mere law, a victory for human rights over international law. At least, that's one side of the story.

Others - and they above all include the addressees of this Humanitarian Intervention - regard this, NATO's first war, as something else: as a relapse into the barbarism which the so-called civilised states have at least tried to avoid since the Peace of Westphalia, albeit not completely successfully (and certainly not in the 20th century).

Morality is the most effective means of justification for wars, which doubtless makes it one of the most important instruments of warfare. There can be no military morality without 'proper' morality as the rearguard.

But what's this got to do with ethics?

There are only few cases where the practical relevance of ethics (a philosophical discipline which is equally interested in theoretical principles and critical application) can be shown more clearly. Morality is the most important part of all our military decisions and propagations; morality is the software deciding for or against war. Cruise missiles can be stopped. But once the morality program has gathered steam, halting it is almost impossible.

We, too, are concerned by the questions raised by this war and the justification propagated - as philosophers, but perhaps also otherwise. The questions include the following: What makes an intervention humanitarian? What are the moral assessment criteria for such an intervention? What strength do these criteria have compared to the pertinent regulations and considerations of international law? What are the specific moral problems of humanitarian intervention compared to the common problems of fighting between different states? And what moral lessons do we learn today from that war? (Please don't raise your hopes too much; I won't be able to answer all these questions now!)

The following thoughts carry on from my paper Is this war good? from the last few weeks of the 1999's war. The first section sums up the essence of that paper.

PART I

1. Self-defence and emergency assistance

1.1 Let's start by taking the customary tack of self-defence and assistance in an emergency. If someone makes an attempt on my life and I can't ward off his attack in any other way, I may defend myself by killing him before he kills me. Note the use of 'may' - I am under no compulsion to do so. I might not value my life so much that I'm even prepared to kill in order to save it. Self-defence is a right, not an obligation.

By contrast, whenever emergency assistance is concerned, it's not my life which is at stake but that of at least one other person. Say a murderer wants to kill a defenceless child. If there's no other way of saving the child's life, may I try and kill the murderer? Of course I may. Indeed, it's probably my duty to do so. Although I may choose to relinquish my own life, I might not be able to morally refrain from saving the child's. In other words, we have a right to self-defence in an emergency, and we may even be obliged to provide emergency assistance. Of course, the scope of actions obligatory for or expected of me when providing emergency aid is not unlimited. It depends on my own personal risk, as well as on whether I am duty-bound to take certain action (e.g. if I am a policeman or lifeguard). Moreover, not all possible actions are permitted in either self-defence or emergency assistance. We'll return shortly to the limits of what is allowed.

1.2 This approach via self-defence and emergency assistance is almost always taken when weighing up the moral justification of a licence or even a duty to kill. It is also used if war is involved. After all, the general assumption goes, states are individuals, too. And any individual, be it a single person or a collective of people organised in the form of a state, may defend its existence - even if this might mean the end of the attacker. Wars of defence are nothing more than cases of a state's self-defence, and wars of assistance (regardless of whether they are fought within the framework of a defence pact) are nothing more than cases of emergency aid. Hence, according to the main argument, they are morally justified. And consequently, as far as the right to enter wars (the jus ad bellum) is concerned, these wars are described as "just wars". So far, so good - perhaps.

1.3 But at this point a problem crops up. States themselves consist of individuals and thus of groups of individuals. Although the chief aim of a state is supposed to be to protect its citizens, not every state actually serves this purpose. What about cases in which the state apparatus turns against its own citizens, or usually of course against individual groups? Do they, too, have a right to self-defence if their very existence is threatened? Of course they do. This is the famous right of resistance - a moral right which the threatened group has vis-à-vis its own state, even if such a right is not enshrined in the state's laws or is even explicitly ruled out. Consequently, external parties have the right to provide emergency assistance in such cases, too, if the group under threat is unable to help itself.

1.4 However, groups - such as political parties, ethnic or religious groups etc. - may be threatened not only by states but also by other groups. The respective state is responsible for countering such threats in line with its main aim, its role of protector. Yet this role still leaves much to be desired in some cases; moreover, the repression, expulsion or destruction of one group by at least one other group sometimes suits those in power, and as well as concealing it they might even encourage or initiate it. In this case, too, if the state ignores its responsibility, outsiders may come to the aid of those not sufficiently able to defend themselves.

1.5 Let me ask you a question: Have you accepted the justification of self-defence and emergency assistance so far without any qualms? If so, you have already crossed a critical frontier; that of the state concerned. Those who concur with the principle that we may also come to the aid of a population under threat on another state's territory evidently believe that the provision of assistance itself is more important - more important than the source of this help, be it domestic or foreign.

And rightly so. If Hitler had not embarked upon foreign conquest and had limited his concentration camps to Germany, ought the rest of the world to have stood idly by because his policy of extermination was kept local? Well, the world might have done so. But under no circumstances would this have been acceptable. (And now comes the sentence which in my 1999 speech deeply affected some people:) This is the point where pacifism becomes a crime. The entreaty "No more Auschwitz!" may well outweigh "No more war". If a second Auschwitz can be prevented, it must be prevented, regardless of its location. Put more generally, violations of human rights are not domestic affairs. Compared to the violation of human rights, violating national borders is the lesser evil, and in fact no evil at all given violations on the scale of Auschwitz. State sovereignty is not the highest good.

1.6 The quintessence is that even interventions by external states can be justified - as long as they are genuine interventions for the purpose of emergency assistance. Such interventions are always described by those carrying them out as "Humanitarian Interventions". If we adopt this linguistic use, the quintessence of this study of emergency assistance is nothing more than:

CENTRAL Humanitarian Interventions can be morally justified.

This is the central axiom of all interventionists. This is hardly surprising, for "Interventionists" is the term given to those who subscribe to this axiom. I am one of them, for despite all the ifs and buts - for which there is no room here - I believe the entire process of justification outlined so far to be basically correct.

An interventionist philosophy is also advocated by the USA and - under its leadership - NATO. It is also subscribed to by other states and organisations, including the United Nations.

1.7 The differences between the USA and NATO etc. on the one hand and the UN on the other concern at most the question of whether the two following intervention axioms - which together characterise strong interventionism axioms - should apply:

HR Human rights can count more than states' rights of sovereignty.
No UN Humanitarian Interventions may if necessary also be launched without the approval of the Security Council or any other UN mandate.

As my crash course on the justification of emergency assistance showed, I too am a strong interventionist.

1.8 As already mentioned, these axioms belong to the very core of NATO's moral software. This software was also deployed during NATO's intervention in Serbia and Yugoslavia during last year's Kosovo-war. Does that mean that this war was and is morally legitimised?

No. These axioms are not a blank cheque. Even Humanitarian Interventions are subject to the same restrictions which apply to emergency assistance. Any moral assessment condemnation of Humanitarian Interventions largely depends on whether at least these conditions of emergency assistance are met. In order for a Humanitarian Intervention to be justified, not only the criterion of ius ad bellum must be met (i.e. in this type of war there must exist a corresponding emergency situation justifying intervention), but also the criteria for the justifiability of the particular way in which this emergency assistance is carried out. Put classically, the relevant criteria of ius ad bellum must also be met for this type of war. Exactly what they are and how last year's NATO intervention appears in the light of these criteria are the questions we will now address.

2.The concept of Humanitarian Interventions

2.1 The range of actions which can be described as interventions is extremely broad. In the most common usage, interventions are nothing more than actions in which the actor intervenes in a process in order to modify or prevent it. In a nutshell, interventions are hence instrumental actions - actions with which the actor aims to achieve certain goals. "Intervention" is thus not an expression of success. In order to be interventions, such actions need not necessarily achieve their aim. They are defined via the existence of corresponding intentions.

2.2 Humanitarian Interventions, in the broadest possible sense again, are actions with humanitarian aims. A multitude of things can be interpreted as humanitarian aims. And the means more or less suitable for such aims are equally diverse, ranging from displays of affection, nursing, a minimum of social assistance, and humanitarian aid missions by the army to a total ban on cars - at least potentially, of course. And just as diverse are the groups of Humanitarian Intervention addressees, i.e. those whom actions of Humanitarian Intervention are designed to benefit.

2.3 The Humanitarian Interventions involved here concern assistance for people who are in an emergency - in situations in which they are no longer able to help themselves. And not just individuals in emergencies, such as a mountaineer who has fallen into a glacial crevice and can't get out by himself. No, what we're talking about is the emergency of many, very many people, and what is at stake is the survival of entire sections of the population - for example when the population of a village is cut off from the rest of the world by flooding and would starve without external assistance. Such interventions are the paradigms of Humanitarian Interventions. How lucky we are if in such cases there are enough military helicopters stationed nearby to carry out rescue missions.

2.4 Let's now be a little more specific. We're not talking simply about people in need, but about self-defence. People are threatened and are unable to defend themselves against this threat. Moreover, this threat is not targeted against individuals, but against an entire group. Furthermore, the type and extent of this threat are extreme. The characteristic aim of the interventions discussed today is this: the protection of groups whose members are already the victims of massive and systematic crimes against humanity, and who will remain so without outside help.

2.5 In the context of NATO and other military organisations, the term humanitarian intervention applies to something much narrower. Humanitarian Interventions are characterised by the following conditions:

  1. The aim of an intervening X is to prevent, end or at least reduce massive systematic violations of human rights inflicted on members of the group Z on the territory of the state of Y.
  2. Actor X: a state or a group of states (e.g. NATO)
  3. Action: military deployment including combat mission
  4. Foreign-state proviso: intervened state Y /= X

Human interventions as defined by NATO are hence not only defined by their aim or the group of intervention addressees envisaged in this aim, but rather by considerable restrictions to the scope of the potential actor (the intervention subject) and the method of intervention (the action) plus an additional proviso. Humanitarian Interventions are thus now limited to military combat missions by one or more states on the territory of another state.

2.6 It should be obvious what this means: from now on, we're talking war. Those who are in favour of Humanitarian Interventions defined in this way are in favour of war. (This is just to clarify the issue; it shouldn't be seen as condemnation!) What distinguishes human interventions from other wars is the wars' particular aim.

Such humanitarian wars, even if only states are foreseen as the intervening party, are still wars in the two conventionally different forms: wars in which at least two states are involved (interstate wars) and civil wars - albeit only those taking place in another state due to the foreign-state proviso. In the latter case, the intervened state Y (the state where intervention is taking place) need not also be the intervention enemy (i.e. the targets of military deployment). The intervening party may have been formally invited to take action by Y - for instance if the group Z is threatened with destruction not by a state but by another group, and both the threatened group itself and the state Y responsible for defending Z are powerless to ward off this threat. Alternatively, the intervention may at least be tolerated by the intervened state in such a case.

2.7 If a state starts Humanitarian Interventions, it wages war. Merely for analytical reasons, such interventions concern issues of life and death; this is part and parcel of the reason for intervention in the first place. Of course, assuming we are the intervening party, it is other people who are being threatened. But this asymmetry vanishes as soon as for analytical reasons other conditions turn our intervention into war. At this point our lives, or at least those of our soldiers, are at risk. [2] At least, this has always been the case so far.

2.8 So much for semantics. Humanitarian Interventions in the above-defined sense (cf. 2.5) are now dubbed Humanitarian Military Interventions, or to be clearer, Humanitarian Intervention Wars [3], just to emphasise that we're dealing with more than just humanitarian military missions to supply earthquake victims or other Red Cross military operations.

3. Humanitarian Military Interventions - the moral restrictions

The list of the criteria which need to be met to legitimise Humanitarian Interventions is as follows. [4]

A HMI is only allowed/justified if:

(i)
  1. The aim is to end etc. massive crimes against humanity (>= "Kosovo Dimension");
  2. this cannot be achieved in any other way
(ii) The manner of intervention:
  1. serves the purpose of intervention;
  2. minimises harm to third parties;
  3. minimises the harm or risk to the intervening agents themselves;
  4. enables the aim of intervention to be achieved with the lowest possible harm to the intervention precipitator
(iii) The intervention itself does not involve massive crimes against humanity
(?-iv-?) The intervention is sanctioned by
  1. international law and
  2. in particular a resolution by the UN Security Council

Criterion (iii) is only included by way clarification; it ought to follow automatically from the other criteria, especially (ii) (b) and (d). The question marks in criterion (?-iv-?) are justified, since this critoerion contradicts the "No UN" axiom of strong interventionism.

PART II

4. General things first

4.1 Moral assessment of emergency assistance or intervention only arises when the above list of criteria is considered as a whole. Just what the 'minimal damage' classified in (ii) into three types (harm to third parties, harm to the intervention subject, and harm to the intervention precipitator) actually means can only be defined more closely by taking other aspects into account. In particular, we will be unable to evade morally weighting these different groups. Morally speaking, does minimising harm to third parties take priority over minimising harm to the those precipitating intervention? And also priority over minimising harm to the intervention subject? It certainly cannot be said for example that the less harm suffered by the intervening party itself (at the cost of third parties), the morally better it is. (This is especially important in our concrete case of intervention.)

4.2 Humanitarian military interventions are terribly complex actions or activities. 'Intervention' covers a multitude of things. It can be used to describe the start of an intervention, the manner of an intervention (e.g. air raids as opposed to ground troops), the same thing but in more detail (e.g. aerial bombing from a greater height), and the manner of intervention used in the various phases of intervention, etc. These are all very different kettles of fish. Consequently, moral judgment of these different things may also be different. Judgments of an intervention as such, i.e. judgments of NATO's Kosovo-intervention as a whole, encompass something terribly complex. It is inevitable that different judgments will attribute very different importance to the different aspects involved.

4.3 The number of subjects/actors is also terribly high in a Humanitarian Military Intervention. In NATO's intervention they included for instance NATO itself, the member states of NATO involved (i.e. the USA, the United Kingdom, France, in particular Germany), the North Atlantic Council, NATO's Secretary General (who ultimately gave the order to attack), the various war ministries, some national parliaments, the population of western Germany (which at the beginning of the war was 90 per cent in favour of it), the population of Greece (which was 90 per cent against it), etc. This is hence a thorny model for all those interested in the topic of Collective Intentionality & Agency.

4.4 Moreover, for each of these levels (of both actions and actors) we must distinguish as strictly as possible between the three classical evaluation viewpoints ex ante, in actu and ex post. Simultaneously exactly what is being evaluated must also be clearly stated: are we talking about an actor, their contribution to the intervention, a certain action aspect, the consequences of a certain action or those of the entire strategy? You can see what I'm driving at. These last few remarks might sound exactly like the start of what in our circles is referred to as the outline of an entire research programme - a programme which in this case would equate to at least three SFB's funded by the German Research Council (DFG). Hence your expectations regarding the rest of my reflections should be correspondingly 'large'. Like everyone else who dares to speak out on this topic, I will endeavour to achieve a terribly reduced degree of simplicity.

4.5 My first reduction is as follows: of the legitimacy conditions in (ii) - the ius in bello criteria - I shall hardly deal with criterion (c), i. e., minimisation of harm to the intervening agent. There's not much to be said about this, since during NATO's intervention this was the criterion which was best fulfilled; NATO suffered no losses whatsoever. (An attacker suffering zero loss was unprecedented in military history anywhere in the world.) If this was the only criterion which mattered, the NATO intervention would be the perfect example of a legtimised Humanitarian Military Intervention par excellence. And perhaps that's just what it is for some people.

By the way, note that this assessment only works if harm or damage to the intervening agent solely means the loss of human life.

4.6 Criterion (ii) (d), according to which those against whom emergency assistance and in particular intervention for this purpose are directed must not suffer more harm than necessary also requires little explanation at this point. It, too, was largely met by NATO's intervention - at least as far as what according to official NATO declarations were the two main precipitators of intervention: the Milosevic regime and his military apparatus (including the paramilitary and various volunteers' groups). The military apparatus, emerged from the whole action relatively intact; the regime itself was actually strengthened by the intervention. (Seen ex ante, wasn't this inevitable? Every group gathers around its leader in times of attack. There are even said to be leaders who know how to take this into account in their calculations.)

4.7 Before going into the other conditions in (ii), let us first look at (i), which could be termed the Intervention Presupposition. This really does deserve though examination.

5. Criterion (i)

5.1 A Humanitarian Military Intervention is only allowed/justified if:

(i)
  1. The aim is to end etc. massive crimes against humanity (>= "Kosovo Dimension");
  2. this cannot be achieved in any other way

5.2 This criterion specifies the very much more general requirement that in order for an action of emergency assistance to be allowed, it must involve a genuine case of emergency assistance, i.e. that the presupposition of such assistance is fulfilled. This means firstly that a situation of emergency assistance must exist, and secondly that the actor must perceive the situation as such (i.e. as a situation of emergency assistance). This difference is important. If the actor perceives - incorrectly - the situation as one of emergency assistance, the 'aid' provided, whatever it is, will not in fact be emergency assistance, but at best supposed emergency assistance. Supposed assistance may be excusable (albeit only under very strict conditions, e.g. concerning the agent's information obligations); it's certainly not legitimate.

5.3 This dual aspect - the intention of emergency assistance on the one hand and the facts of emergency assistance on the other - are shared by all interventions, including humanitarian and military ones. But only regarding their evaluation, not in conceptual terms. (Perhaps this, too, is something which makes interventionist self-justification so natural.)

Let us take for example 'deterrence interventions'. These are simply actions with which the deterrence subject intends to deter somebody - period. The mere existence of such an intention is enough to make an action one of deterrence. In other words, deterrence actions in this broad sense are nothing more than attempts at deterrence - without any presupposition of facts. Such types of deterrence interventions hence have a thoroughly subjective touch. They become such interventions simply when they are perceived as such by the intervention subjects. We know of many deterrence actions in which there was no reason justifying deterrence in the first place.

This is exactly the same in conceptual terms for Humanitarian Interventions. An intervention is humanitarian if it is associated with a humanitarian intention. Whether the state of emergency assistance supposed by the subject actually exists is a different matter. Humanitarian Interventions being launched without the elements of emergency assistance existing is not a conceptual impossibility. It may occasionally not even be a real one.

But let us return from this little conceptual digression to ethics.

5.4 In order to be morally justified, Humanitarian Interventions - now in parallel again to the general emergency assistance analogy - also have to correspond in practical terms to their self-imposed image of emergency assistance. In other words, the group on whose behalf the intervention has been undertaken must actually be in a situation of emergency.

But exactly what sort of emergency? The criteria specified in the relevant literature vary enormously. This is partly due to the fact that what I call Humanitarian Intervention Wars are rarely sharply defined and are hardly ever called as such by name. It is generally believed that the relevant reasons for emergency intervention should be violations of human rights. As globalisation proceeds, Humanitarian Interventions should (according to common opinion) be developed into an increasingly accurate instrument of universal human rights policy.

However, it is usually acknowledged that not every violation of human rights is an acceptable reason for intervention - not even those carried out systematically by a country. Otherwise, a war of humanitarian intervention would even have to be carried out against the USA, for according to Amnesty International, capital punishment (which is practised in the USA) is a violation of human rights. We, too, would be at risk: after all, many people outside the Western world fail to understand why in our society hundreds of thousands of people to whom we owe the most are kept in isolation in 'special isolation centres' as soon as they have reached the age of 80 and require nursing care. Hence the oft-used definition of a reason for definition as a "reason for people's abhorrence" might not be so good after all - not even for us. And I am assuming that the definition sought cannot be a culturally related one.

The definition of a reason for intervention put forward by one of the most commonly quoted writers on Humanitarian Intervention theory as a situation inviting "transboundary [forcible] ... help, provided by governments to individuals in another state who are being denied basic human rights and who themselves would be rationally willing to revolt against their oppressive governments" (Fernando R. Tesón: Humanitarian Intervention: An Inquiry into Law and Morality, p. 5) is probably also unusable since it declares the overwhelming majority of states in the world to be precipitators of intervention.

The question is, how serious must violations of human rights be in order to justify intervention?

5.5 In my 1999 paper on the Kosovo-war, I simply ignored this question. The "Kosovo Dimension" was sufficient, ran my answer. By this I meant the scale of those crimes against humanity "which we were made to believe caused NATO to intervene in the Kosovo crisis", i.e. all the massacres, organized rape, mass expulsions etc. listed by the USA and other states as the reasons for their intervention. Let us refer to the scale of these reported crimes in the following as the Kosovo Dimension (which is already referred to in criterion (i) (a)).

A HMI is only allowed/justified if:

(i) (a)The aim is to end etc. massive crimes against humanity (>= "Kosovo Dimension")

5.6 I believe that last in 1999 many other interventionists used the same "Kosovo Dimension is sufficient" solution to circumvent the issue. That this is no longer possible, as a number of questions of general importance and on the actual facts have since cropped up. Let us start off with the general questions, or at least with some questions which are more general.

The question is not whether the thus defined Kosovo Dimension corresponded to the facts. I shall simply continue to assume this. The two most important questions of general intervention assessment relevance are then:

(a) Is criterion (i) (a) perhaps too strong? Must massive crimes really assume this Kosovo Dimension before intervention can take place?

And vice versa:

(b) Was the willingness of NATO (and many of us) to intervene somewhat premature? Was the "Kosovo Dimension" really enough to cross the critical threshold necessary for Humanitarian Military Intervention to be be obligatory / permissible? In other words, is criterion (i) (a) perhaps too weak?

Even if formulated clearly, these questions are ticklish. This is why they are almost never asked. Nevertheless, they have to be asked, since the lives of thousands of people on all sides involved depend on the answer (unless of course NATO's high-tech is used). Can we simply skirt around clarifying these questions?

At exactly what point do violations of human rights become so bad that intervention is obligatory? And if you're not keen on the word 'exact' here and would rather avoid precisely determining this grey area, at what point is intervention unquestionably required? (At this point you will, as some of you will already have realised, have simply repeated the initial question, but this only serves to make the two questions appear all the more urgent.)

5.7 At this point we should mention something which perhaps should already have been said a while ago. Regardless of where the intervention threshold is to be drawn, the scale of the crimes addressed by these considerations is such that our question appears inappropriate. At what point are crimes against human life and human dignity so bad that action may be taken against the criminals responsible? This sounds as if whether the opposite might also be allowed were a moot question. But this is not and must not be allowed to be the case.

Hence the following should now apply to Humanitarian Intervention Wars: they are only allowed if they are also obligatory; in a nutshell, if being allowed and being obligatory coincide.

(A=O) HMIs are allowed iff they are morally obligatory.

5.8 Note that no formal problems result from this (A=O) postulate. After all, the equivalence:

(A=O)* Allowed (p) = Obligatory (p)

can always be made true by means of interpretation structures in which there is exactly one deontically perfect world. In such a structure, the following holds: If X holds in least one deontically perfect world (X is allowed), X holds in all deontically perfect worlds (X is thus obligatory) - and vice versa.

5.9 By contrast, the (A=O) postulate may prove problematic in terms of content. One objection could run as follows: Assume two exclusive alternatives X1 and X2 are equally good. Since they cannot both be implemented, both X1 and X2 cannot be obligatoy. However, owing to (A=O) neither of them would be allowed. But this (runs the objection) is plainly counter-intuitive. There is doubtless something in this general objection geared towards the structure of the principle. However, it fails to take into account that if a Humanitarian Intervention War has two equally good alternatives, this war is not the last possible means (hence violating (i) (b)) - and is thus neither obligatory nor 'merely' allowed. The "possibly two equally good alternatives" premise of this objection is not fulfilled for allowed/obligatory Humanitarian Intervention Wars.

5.10 Although the (A=O) postulate doesn't help particularise the intervention threshold, it still delivers something important: namely, a metacriterion for the adequacy of each such particularisation. One conclusion of the postulate is namely:

HMIs are only allowed if the violations of human rights to be ended are so appalling that a military intervention as response to them is obligatory.

Note, however, that this doesn't rescind the other intervention presupposition (i) (b), according to which such an intervention may only be the ultima ratio. It thus continues to hold that a Humanitarian Military Intervention is not allowed if other permissible alternatives exist.

5.11 However, let's not avoid even tortuous problems (even if analytica philosophers are especially inclined to do so). How awful do crimes have to be for war to be obligatory as the ultima ratio?

This is a question which has been avoided by more than just analytical and other good philosophers. It has also been sidestepped by those who decide over whether to fight or not to fight, or (to take into account the German contribution in 1999, and also possibly the year before) those who share responsibility for such decisions.

Everyone knows the easiest way of rhetorically concealing such evasions. Instead of answering the difficult question, you answer a much simpler one with the strongest possible answer imaginable. Let me repeat the question it all boils down to: How bad must crimes be for war to be obligatory as the last possible means? The ultimate answer in 1999 was "Auschwitz". As a paradigm for a humanitarian worst case in which the question of war is anyway only hypothetical, this answer would have been correct. (This is precisely how I myself argued in my first quick attempt at justification; see § 1.5 above.) Yet as threshold justification for NATO attack? In this case, the attack should definitely never have taken place. However, this example of Auschwitz instrumentalisation wasn't meant quite like that.

Claude Lanzmann, the French director of the film Shoah, put this very aptly in an interview: "These ... references to the Holocaust are a way of muzzling all discussion. Talking forbidden! Argument over!" (Tariq Ali (ed.), Masters of the Universe? London/New York, 2000, p. xviii.)

5.12 Paradigms for justified humanitarian intervention include India's 1971 intervention in what was then East Pakistan (to put an end to the genocide of the Bengalis) and Tanzania's intervention in Uganda in 1979, which ended the massacres carried out by Idi Amin. However, with respect to the scales of the massacres which preceded them, both these interventions exceed the maximum of the "Kosovo Dimension" victims 100 times over. Moreover, India's intervention in East Pakistan is also the paradigm of justified humanitarian intervention without a mandate from the UN. By the way, the state which protested most sharply in this case against the lack of a UN mandate was the USA.

5.13 We will have to carry on tormenting ourselves for some considerable time before we can find answers to our moot questions which go beyond ad hoc responses. What the aforementioned research project would have to achieve is obvious: it amounts to nothing less than drawing up a scale classifying types of human rights violations by varying severity: the nightmare of the Holocaust, the genocide of the Armenians, expulsions, rape camps, torture, Srebrenica, children's transports, mutilations, etc. etc. - the almost endless list of all the dreadful things mankind is capable of doing.

5.14 However, even drawing up a scale of Homo sapiens' negative potential of species still fails to address a classification scale for the intervention obligation. What factors need to be taken into account when weighting the intervention obligation borne by outsiders? Other things being equal (for example assuming the same military potential), is it those who are closest who are most obliged to intervene? But does 'closest' mean geographically closest, or members of the same culture or religion? A nation who speaks the same language? Or should all intervention armies be mixed to prevent further escalation? And should the members of such campaigns be restricted to volunteers, or may conscripts also be sent? Or alternatively, are only the very best fighters, usually mercenaries, good enough for these humanitarian purposes?

5.15 The most important question of all is who should decide all this? When it comes to asserting universal questions of human rights by means of military force, what sort of decision-making procedures are permissible? Should all these questions only be left up to Washington, Moscow and Beijing? Or perhaps the organisations of states of the continent concerned, e.g. the OSCE?

5.16 And what needs to be done to prevent something which totally contradicts the basic universalistic idea of the moral/intellectual software thus conceived for these 'human rights wars' being built into them in the first place, namely the chip which sets off the great mutual Humanitarian Intervention War, the 'clash of cultures'?

5.17 These and similar questions will keep philosophers busy over the coming decades if humanitarian human rights wars become the strategy of the future. They are at any rate the Pentagon's future strategy. The recent attempts at restructuring of NATO and the German Armed Forces follow the same concept. Now that NATO and the German Armed Forces have officially abandoned their strict defence character without even a hint of public discussion, another semantic correction would soon be in order: ministries of defence would have to be renamed ministries of intervention. I mean ministries of humanitarian intervention, of course.

If this is to be the future, you don't need to be a prophet to predict that the need for humanitarian intervention experts will spawn a growth sector in practical philosophy. In fact this need already exists; this is one of my core sentences today.

5.18 But let us return to the present. Was NATO's Kovosov-intervention okay? As far as this question is concerned, we are still dealing with the first step, with criterion (i) (a) - the humanitarian intervention presupposition. And to be honest, we're not making much headway.

So let's ask again. Assuming there really had been no alternative, given the Kosovo Dimension, was NATO's intervention really obligatory/permissible?

6. The "Kosovo Dimension" vs. the Kosovo Dimension

6.1 To be able to tackle this question, we need to know one thing right from the start: how big was the "Kosovo Dimension"? What exactly does it comprise? Assuming it really was this factor which tipped the scales, what prompted former Spanish anti-NATO activist Javier Solana on 24 March 1999 in the role of NATO's Secretary General to order the attack on Yugoslavia?

This may not be an easy question to answer. Nevertheless, it's still the easiest of my long list of questions. As we have already seen, not everything counts which may have been decisive: only the humanitarian aspect is relevant for determining the "Kosovo Dimension" at the time - in other words the essence of the reports which then (and now) filled us with such revulsion. This dimension is easy to verify, since the reports still exist.

6.2 Assuming these reports were correct, were we really obliged to attack? The attitude "We cannot simply stand by and watch" is a matter of course in reaction to such humanitarian disasters. But this by no means justifies war - not by a long chalk. However, we'll not change the subject like everyone else seems to, but repeat the question: With the benefit of hindsight, was the "Kosovo Dimension" on a scale justifying Humanitarian Interventions? In other words, applying the universilisability postulate and allowed-iff-obligatory postulate, something which - other things being equal - would need to be done in other cases on the same scale?

Let's assume it was. This would then have had consequences - including (to name just one of the weakest ones) this: Supposing in all cases in which a) the "Kosovo Dimension" was also reached and possibly greatly exceeded, b) we were in no lesser a position to intervene, c) this intervention did not require a Humanitarian Intervention War or any other type of military campaign whatsoever, and d) all other emergency assistance legitimacy conditions (with the exception of (ii) (d)) would easily be met anyway - where would we be in such a case? At the very least forced to provide some justification. You all know that such cases are far from just hypothetical. Just think of Turkish Kurdistan. What are the arguments here? And do these arguments stand up to closer analysis?

The question is not one of possible explanations. A number are already emerging. It boils down to justifying why certain differences between these two cases are morally relevant. But how can they be justified?

6.3 One of the many explanations is trivial: for NATO to launch a war of Humanitarian Intervention against Turkey would be conceptually impossible - this follows from the interventionist foreign-state proviso. But wouldn't this be another argument for initiating pro-Kurdish Human Interventions? Interventions which, as we have already said, are a) (for a number of years) at least relevant and necessary compared to the Kosovo Dimension, b) are incomparably easier to carry out, c) would not involve any military action whatsoever on our part, and d) would optimally meet all claims of legitimacy required for emergency aid.

There are plenty of other similar cases. The US president could have relaxed the economic boycott against Iraq at a stroke - and the lives of some 300,000 children would have been saved.

6.4 This inconsistency argument does not argue per se against the legitimacy of NATO's Kosovo intervention. If intervention is really obligatory at three different places, it is always better to carry out intervention at one place at least rather than none at all. But this raises a new problem of justification. What factors argued ex ante for intervening at the place where intervention was the most risky? How does this disappear ex post? What rule justifies these differences? Are interventions more moral the riskier they are - riskier for everyone up to the point of jeopardising world peace? Do military interventions really always take priority?

We're not stupid. We know that international politics involves a wealth of other factors alongside morality. There are economic interests, superpower ambitions and forces, exactly timed diversions of attention, and thousands of other factors. The only crucial question is whether all these things together are allowed to make a difference? The moral difference between a stroke of the pen and 37,000 air raids?

6.5 How bad was the "Kosovo Dimension"? The cause of military intervention, does it still stand up to our moral analysis? After all, the "Kosovo Dimension" always meant the dimension we were made to believe by the media to be genuine.

But was it? This is certainly not irrelevant for the ex post assessment of the NATO intervention. How can it be verified? What channels can be used to find out - CNN, ZDF, TV Belgrade, UCK pirate broadcasters, Zet-Net, Human Rights Watch? It quickly emerges that our entire philosophical problems with verification and falsification, testing hypotheses, coherence presumptions and whatever geared towards the scientific community are a piece of cake compared to those encountered when we try to find out the 'truth' about the actual dimensions of our real world. These are thus problems which at the moment are even too big for me.

But there is one area in which nearly all the serious sources agree. As far as the number of dead is concerned, the actual Kosovo Dimension was about a tenth of the "Kosovo Dimension" portrayed in the media. This indicates that when we are considering Humanitarian Intervention War, we should be at least three times more sceptical than we were in 1999. But as we know from experience, this advice has never yet been followed.

Are there any more questions regarding criterion (i) (a)? Of the above-mentioned three major DFG-research programmes, this criterion would occupy a whole programme by itself. The main question "Was the Kosovo Dimension a reason for an Intervention War?" will doubtless continue to dog us for a while yet.

What about the other criteria? Re: (i) (b): Alternatives? Permissible alternatives to do what? To carry out air raids? Certainly. To carry out any form of military intervention? I don't know.

7. Criterion (ii) (a)

A HMI is only allowed/justified if

(ii) the manner of intervention (a) serves the purpose of intervention

7.1 The purpose of intervention? Well, any action can have many different aims. And the action concerned may help achieve some of these aims but not the others. This also holds for interventions, irrespective of whether they are humanitarian. The aim of an action usually doesn't exist, unless one means either the overall goal of the action, i.e. the sum of all the aims, or the primary goal, i.e. the reason for which the actor performs the action in the first place, with the other aims for their part only being means to reach this primary aim.

7.2 Criterion (ii) (a) focuses exclusively on the aim which makes the Intervention War a Humanitarian Intervention - the prevention of those massive violations of human rights whose problems of scale we have just tackled. And this goal need not be the same as the overall aim of the intervention - for why should not such a huge action (with so many actors involved) also be combined with other aims? And this aim needs not even to be the primary one.

7.3 In a Humanitarian Intervention War, must the humanitarian aim also be the primary one in order for such an intervention to be morally okay? No, it need not. This is not even required of simpler aid actions - at least when we are assessing the action, not the actor. Suppose a child is locked in a burning house. If a somebody rescues the child, even though their sole aim is to become a hero, such supplementary aims do not worry us as long as the child is rescued, even if they were primary aims for the rescuer. Rescue, whatever the motives, is still better than no rescue.

With respect to the NATO intervention, contrary to the popular game of speculation about all the possible aims and motives possibly associated with it, I will simply have to leave them aside. Moreover, speculation simply diverts attention away from the central questions.

7.4 And they include, even though it still belongs to the preliminary clarification, this: Supposing an intervention had certain positive effects which had not originally been intended? The response is clear: although these consequences cannot be attributed to the actor, they remain associated with the action, which in this respect was also positive. Hence, even if it is difficult to imagine, suppose (rather counterfactually) that NATO had bombed bridges in the Vojvodina, i.e. far in the north of the rest of Yugoslavia, and that these bombardments in the north had actually stopped (in some way or other) the expulsions in Kosovo right in the south of the country, yet NATO's aim was not the latter, but rather - well, it doesn't matter what it was. How should the intervention then be assessed with respect to these humanitarian consequences (and for the time being simply ignoring all the others)? This is no different from the case of a child trapped in a burning house who is only saved because a burglar who just broken in had managed to crack the electronic locks preventing the child from escaping.

7.5 Was NATO's humanitarian aerial warfare against Yugoslavia useful to its (at least thus declared) humanitarian aims? Did the bombardment of Yugoslavia day and night for 78 days minimise the "Kosovo Dimension"?

No. As far as the humanitarian aim of the intervention is concerned, the humanitarian bombs were solely counterproductive. They couldn't have been anything else. This must have been clear to all the experts right from the very start. And it was clear to the military experts. This type of intervention was counter-productive for two reasons.

Firstly, without the bombardment, the floods of refugees would not have occurred on the scale which we came to read about as the Kosovo Dimension reaching over a million people once the bombing was over. For hundreds of thousands of people in Kosovo, it was the start of bombing itself which caused them to flee their villages and across borders, and it was only when bombardment started (or to be more accurate when the OSCE observers began to be withdrawn and bombardment was announced) that the Serbian attacks escalated. This had been predictable; this escalation was for example forecast by top 'experts' such as Wesley Clark, NATO's Supreme Allied Commander Europe with a master's degree in Philosophy, Politics and Economics from Oxford University. By bombing Yugoslavia, NATO was pouring oil on the fire in Kosovo which had by no means been stirred up by the Serbs alone.

7.6 Secondly, the bombardments did not just directly contribute to the escalation of the cleansing directed against the Albanian population and hence increase rather than reduce the number of expulsions; they also helped to bring about what the intervention was allegedly supposed to prevent, namely further ethnic cleansing. Nowadays, thanks to the bombardments and the victory of the UCK they brought about, ethnic cleansing in Kosovo is largely complete - albeit the other way round. Therefore the opinion that bombardment prevented further violations of human rights is incorrect.

7.7 Nowadays, the then strategists proudly report that most of the Albanian refugees driven out of Kosovo could since have returned home. This, thank goodness, is true. But it is not an argument justifying (by way of necessity or usefulness to the aim) the bombardment. The refugees and expellees have not been 'bombed back' to their homes; it was only the cessation of bombardments and not the bombardments themselves which enabled them to return.

The air raids were not just the worst conceivable means of protection against murder and expulsion; they were totally unsuitable for bringing about this protection.

7.8 This assessment contains many premises I cannot vouch for. Too much depends on the uncertainties regarding the factual Kosovo Dimension, rather than that presented in the media. And at least as much again depends on something which previously has not been mentioned at all - on the reasons and the historical, religious and social background to the whole affair. Consider on account of all these premises my assessment to be merely something like my personal opinion. But do let me know if you have good reasons for a different assessment.

My assessment still leads me to a clear conclusion. The NATO war violated criterion (ii) (a) -it was neither permissible ex ante nor justifiable ex post given the deterioration of the situation thus caused.

7.9 NATO's air raids affected not only Kosovo but also the rest of Yugoslavia. They were not restricted to the military infrastructure. In fact this was affected least of all; the attacks were primarily aimed at destroying the country's infrastructure. And this aim was largely achieved. To what purpose? If I intend to reduce somebody's willingness to commit violence, can I achieve this aim by destroying his basis of life? Let me ask again: To what purpose? To what purpose, mark you, which could be comprehensibly associated with the allegedly primary Humanitarian Intervention aim? And to what purpose, whose necessary attainment was accompanied by the unavoidable side-effect of "collateral damage" amounting to over 500 dead, including about 80 children, into the bargain? Try as I might, I cannot find such a purpose. But perhaps someone can tell me otherwise?

8. Criterion (ii) (b)

A HMI is only allowed/justified if

(ii) (b) The manner of intervention minimises harm to third parties

8.1 Humanitarian Interventions are cases of emergency assistance. This was the beginning which, assuming you followed me, made us interventionists. [5] Following this start, assuming it is strictly orientated towards our starting-point (self-defence and emergency assistance), we wouldn't have to spend too much time on criterion (ii) (b). After all, the relevant section of criminal law states quite clearly who must not be harmed in the case of normal self-defence or emergency assistance: self-defence and emergency assistance may only be directed against the attacker, not third parties or their property. The well-aimed sniper's bullet against the terrorist holding hostages, if this really is the last chance of saving them, may be okay under criminal law, yet ceases to be so as soon as another innocent person is jeopardised by this bullet.

8.2 Although thoroughly acceptable under criminal law, within moral consideration (our activity here) it will be impossible to maintain this prohibition under all circumstances. At this point we ought to embark upon a process of weighing-up similar to that acted out for or against utilitarianism in any introductory seminar course. To borrow one of the most common exercises, let us assume that a terrorist has taken 20 hostages, and let us assume we are all absolutely certain that as his demands have not been met, he will blow himself up together with all the hostages in the next few seconds. Shouldn't the SAS marksman who already has his sights trained on the terrorist be allowed to fire, even if he can't completely rule out the chance of hitting an innocent passer-by who suddenly appears and strays into his line of fire? And if you hesitate, would you do the same if the terrorist had taken 50 hostages? Or what about 200? Or 1,000? These reflection games are terrible. Then again, ethics isn't supposed to be a barrel of laughs.

8.3 The Kosovo Dimension easily outweighs all bank-robber scenarios. Those who in view of this scale of difference accept a Humanitarian Intervention War as I have defined it as a prima-facie option have already made up their minds. For moral reasons they are willing to overstep the bounds of what (in related contexts) is permissible under criminal law. We are thus entering a field where what is forbidden by criminal law is morally allowed. Intervention Wars which do not jeopardise external parties simply do not exist. Even Intervention Wars with the highest of humanitarian intentions are no exception. It's impossible to approve of Humanitarian Interventions and at the same time to rule out others being put at risk.

8.4 This certainly doesn't mean that the greater the horror the Humanitarian Intervention is supposed to combat, the less careful we need to be about putting innocent parties at risk. Students of intervention ethics will have to tackle tortuous questions such as the degree to which a link nevertheless exists between the necessary struggle against the intervention precipitator and the risk to outsiders. Human shields tied to attacking tanks, workers at factories important to the war effort - these are the sorts of cases which need to be examined more closely in connection with this question.

8.5 However, we don't need this discussion for the first NATO war. At least not if you share the view that the air raids were not a means which served the humanitarian purpose - for there was no need for the "collateral damage".

9. Closing remarks

9.1 Even if the actual Kosovo Dimension matched the "Kosovo Dimension" in the media which was accepted by us (at least accepted in spring 1999) as a good reason for intervention - i.e. the causa iusta actually existed, this causa actually corresponded to the intention of the intervening agents, and that in addition there was no genuine alternative - even given all these assumptions, in my view the manner of the intervention contravened the central rules of the ius in bello and must therefore be condemned.

9.2 I hardly paid any attention to questions of international law in my paper. This isn't because these questions are unimportant for intervention ethics. The reason is quite simply that if something is already forbidden for moral reasons, this prohibition is not rescinded by legal legitimisation. Clarifications from the angle of international law are of course highly relevant, but this would sidestep the essence of whether this war was morally justifiable.

9.3 The discussion in Germany about the NATO war was doubtless not analytical. The actual question over whether this war is really morally permissible, i.e. really morally obligatory, has been largely ignored; in order not to ignore it, plenty of other questions (as I have tried to show) should have prompted. These questions have not been shelved. Humanitarian Interventions are still on the agenda of the USA, NATO and other states. We must ask ourselves these questions, even if they hurt. We need to tackle them with the best instruments at our minds' disposal, including analytical philosophy.