« Preventive War - Becker | Main | Response to Comments »



Feed You can follow this conversation by subscribing to the comment feed for this post.


Interesting comparison to invading germany after Hitler took the Rhineland as a justification to preemptive war.

Using that logic if Canada and Mexico were to attack America in a preemptive war because Bush invaded Iraq without cause and that they believed America was on a path to world domination - would that be justified?


A nitpick. The first word of the title is either mispelled or misused.

The content is high class and thought provoking. Thanks.

Paul Eremenko

Society constructs certain rules, the widespread compliance with which allows for a reduction in various costs. The rule of honesty, for instance, might decrease contracting and contract enforcement costs; the rule against unprovoked violence allows people to invest less in personal protection and thus reduces the costs of transacting day-to-day business that involves walking on the streets.

The rule against attacking a nation unless attacked introduces a degree of certainty into international interactions that presumably allows everyone to reduce the cost of self defense. Once this implicit international consensus prohibiting preemptive war is broken, the costs of defense for all nations will likely rise. It is the discounted value of these future defense expenditures that Judge Posner appears to omit from his cost-benefit example. It seems that at least a plausible case can be made that this will not be a negligible term in the cost-benefit balance that might make the threshold at which preemptive action is justified (at least the first time) extraordinarily high.


Dear Justice Posner,
I am a 2L at DePaul and I just wanted to say that I think all of your legal decisions are brilliant. I think that you and Dr. Thomas Sowell are the most insightful economic minds in the world today.


Dear Judge (and hopefully one day Justice) Posner,

I'm a 2L as well (at Georgia) and have had a deep admiration for your judicial opinions, which it seems, have graced almost all of my casebooks so far.

I will comment only that I think that both 1) your Rhineland example is a perfect case of when preemptive war is optimal and 2) the costs far outweigh the benefits of our present venture in Iraq. I would have said so (and did) even from an ex ante perspective. It seems to me that, even where we benefit from attacking Iraq in insolation (a notion I might also dispute) the kick-to-the-hornet's-nest effect this will have on the Arab world will only bring enormous costs to the nation in years to come.



Your analysis is surprisingly one-sided--it seems not to include any of the costs of preventive war.

Consider, for example, the costs of perceived illegitimacy, which tends to follow preventive strikes. The more distant a threat, the more illegitimate the war will appear in the eyes of (a) potential allies and (b) the citizens of the country one invades. In other words, the more distant the threat, the higher the cost of the war.

And, of course, perceived illegitimacy doesn't stop with the war itself. Post-Iraq, the US is seen as less credible and less moral by millions of people, some of whom may be in a position to help us kill terrorists.

There are other costs; why not address them?


I respectfully disagree that the Nazi reoccupation of the Rhineland illustrates the benefits of your model. The British did not attack Germany in response to the reoccupation of the Rhineland because they underestimated the threat posed by Germany. Simply put, the British believed that Hitler would stop at the German border. The French were immobilized by their internal politics, and were loath to fight without British aid. In the real world then, the anticipated costs of attacking Germany were likely lower than the anticipated benefits of such an action. Clearly the French and British governments should have reacted more forcefully to Germany's treaty violations up to and including the reoccupation of the Rhineland. However, it does not seem likely that an economic analysis would have supported invasion of Germany at that point in time.

John Hempton

This preventative war thing is pretty difficult. Alternative histories from 1936 are possible - and the assumption that Hitler would have been overthrown is too quick by far. (We were sure Hussein would be overthrown after Gulf War I.)

But how is this for an alternative history.

Hitler manages to portray himself as a victim. Invasion falls back to hard lines (there is no overwhelming Allied build up then and Hitler had not expended himself in Stalingrad.)

As a result Hitler never looks East until the West is thoroughly beaten. Japan never joins the war - and America thus also sits on the sidelines for way too long.

Hitler then controls Western Europe including Britain but his heart is set on making the German minorities of the East German majorities. The war ends in a cataclysmic struggle between fascism and stalinism.

You are assuming in your post that free market liberalism had wide enough support in countries like France in 1936 to make all that possible. It did not.

Being a collaborator had romantic appeal for some French in 1940. Because free market liberalism has been such a success after WW2 we should not assume that it had a natural following of scale before WW2. We should also not assume then that free market liberalism would win a war in which it could be perceived to be the aggressor.



The probabilty of an attack is only certain, when the attack has occured. If that is the case, then the benefits of a preventative war, is only evident after the fact.

The answer I get from this is that we don't know with any certaintity that premptive strikes have benefits. The benefit is only seen, after the fact. Those who want to divert back to "well, what would happen if we didn't go after Hitler," should play craps in Vegas for a few hours, and see how fast their money evaporates.

Preemptive war like gambling at the craps table is just a bad way of making money. Stick with selling stereos.


One potential difficulty I see with using a cost-benefit analysis to decide whether to engage in a preventive war is that any such analysis seems (at least as a practical matter) to ignore or otherwise be biased against information contained in the other side's revealed "commodification" preferences. By waiting at least for the other side's strike to become "imminent" we gain valuable information about their LACK of willingness to fight that would not otherwise available. Rapidly changing circumstances and the imperfect lines of communication between two parties that hate each other would seem to make such information even more valuable. Although analytically we may be able to account for that in the CBA, as a practical (perhaps even pragmatic) matter I'm not so sure we can do so accurately.

Of course, if the other side has already revealed its preferences, for example, by attacking another state, then we can estimate its commodification preferences far more reliably. Germany's invasion of the Rheinland, and Iraq's invasion of Kuwait, may be examples. But was our most recent invasion of Iraq?


What do you mean by "probability" in this context? Are you trying to say that if the Nazis were to occupy the Rhineland again and again, then as the number of Rhineland-occupations tended toward infinity, then the proportion of them which resulted in war in Europe would tend toward fifty per cent? Presumably not, because that's pretty plainly incoherent.

So I'm presuming that this is a Bayesian concept of "probability as degree of belief". Which rather points up the weakness in this doctrine of pre-emptive attack; unless the attack is imminent, then what you are talking about is the subjective assessment of a policy-maker, with no real evidence to back it up. Since what a policymaker believes is most clearly influenced by what he wants to believe, then what you have here is a doctrine of wars of aggression, with its indecency covered up by a fig-leaf of pre-emption.

What other large and costly government projects would you suggest that this doctrine could be applied to? For example, if President Bush were tomorrow to wake up and decide that the USA needed a National Health Service, massive cash reparations to the descendants of slaves and to take the commanding heights of the economy into public ownership, would you regard it as sufficient reason for doing so that he had subjectively assessed the probability-weighted payoffs from doing so as positive?


The problem with your arguement is that there is no set standard as to determining if a preventitive war will in fact be preventitive of anything and if it the benefits of the war will outweigh the costs. How do you decide which countries to attack and at what point? Which countries pose a threat?

As for Iraq, you must remember that the White House withheld information about North Korea's nuclear program until after a congressional vote deciding our plan of action has taken place. If we were to take your Rhineland example into account we should also admit that both North Korea and Iran pose a much greater threat to the world than Iraq ever did. That is clear. Yet no preventitive measures are being taken now. The worst the current administration has to offer is sanctions. In fact, because of the "preventitive war" in Iraq the Bush administration can not wage a preventitive war in Iran or North Korea. International relations aside, the mere logistics of such an event would stretch our armed forces to a breaking point and would have major negative effects on the economy.

A preventitive war must be planned in good faith and the planning must deal with immediate changes in overall intelligence. Once news of the North Korean nuclear program broke out the preventitive war for Iraq should have been put on hold. A decision should have been made as to who posed a bigger threat. Intelligence concerning WMDs in Iraq was known to be shaky even then where as North Korea outright admitted to possessing weapons of mass destruction.

What you have to understand is that things like "preventitive war" and "preemptive strike" are all terms used to excuse whatever the current policy is. One should never take any such statement at face value. Take the trade embargo with Cuba, for example. The official explanation is that the embargo is still in place to punish a repressive Communist regime. Yet China remains our biggest trading partner and it is the biggest Communist state in the world.

One should never try to rationalize any action without first analyzing whatever data is at hand.


Your simple numerical example, which compares the expected costs of waiting to be attacked, military build up, and preventive war, leaves out one important alternative: waiting to collect more information about the costs and probabilities. Waiting to collect information may turn into waiting to be attacked if your standards for sufficient information are too high, but it is a distinct alternative, and an attractive short-term strategy, in situations where:

1. The threat of imminent attack is low,
2. There is a decent chance of gathering enough information to change which of the other alternatives you consider to be least costly, and
3. The costs of the other alternatives are not likely to increase significantly while you are taking time to gather information.

Given that Iraq was largely incapacitated by various restrictions and that weapons inspectors were in the country learning about its capabilities, Iraq in March 2003 seems to be a strong candidate for meeting these three criteria. Thus, continued inspections, rather than preventive war, would have been the cost-justified course of action.

(technical note: there seem to be problems with posting after previewing)

Sampo Syreeni

As dsquared noted above, talking about probabilities in the context of preventive war is a bit misleading. If the decision to go to war was a mechanical, predictable, well-understood process, we could perhaps assign it an a priori probability and work the expected costs and benefits from there. In this case cost-benefit accounting is both reasonable and efficient.

Take the drunk driver. We understand the consequences of impairment relatively well, so we can assign an expected cost to driving under the influence. In many cases the cost justifies a legal incentive not to drink and drive.

But how about what Knight once called uncertainty, as opposed to quantifiable risk? When we talk about unique, one-time events, we cannot assign a definite probability to them. Activity related to them is entrepreneurial in the Austrian sense, and quite possibly beyond cost-benefit accounting. This is especially relevant when the outcomes depend on people's decisions, as opposed to a mechanical process of one kind or another.

Take the person sitting next to the driver. Suppose she's drunk. Should this be criminal on the grounds that a drunk person's advice might lead to a crash? I'd say no, because ultimately the driver is the one responsible for the safety of the passengers, and there is no clear causal relationship between what the passengers say and what the driver does. For the most part the law agrees -- we regulate the conduct of the proximal actor, because he's the one making the final decision, and we cannot reasonably treat him as an automaton which can be made to operate according a nice statistical description. (OTOH incitement offences provide a notable exception to the reasoning. But should they?)

I think this is generally speaking why there are no thought crimes, but instead punishment only follows definite actions, or at least definite signalling (e.g. declaring war, breaking into a house or building a bomb). We modify people's incentives so that rational judgement of costs and benefits will make crime unprofitable but do not try to touch the mechanisms leading to the final judgment. The same then goes for preventive war.

Of course war is hazier in that it's a collective phenomenon. It is difficult to assign guilt and to put the proper disincentives in place when considerably more than one person participates in a crime. War is also messy in that there's a grave imbalance of cost between building something and waging war on it, and also between causing indiscriminate destruction and punishing only those responsible. Terrorism is even tougher because individual disincentives do not work -- how do you punish a suicide bomber?

But does this mean the basic reasoning is incorrect? Not likely. All it does is raise the potential cost of war. If it's true that the a priori probability of war cannot be reliably estimated, the only reliable disincentive we can give is to commit to responding in kind. Which is probably why mutually assured destruction worked -- even if we didn't know the expected cost, we indirectly knew that going to war would not be profitable.


Given that no theory of preventive war is acceptable, and given that opponents of preventive war will not accept anything short of absolute certainty (dead people and smoking gun in hand of the accused), then the next logical question that I can think of is: how high should costs be allowed to reach before reacting? Of course, both assumptions could be relaxed, but let's run with this ....

Ex: Since it is accepted that Saddam was paying families of suicide killers in Israel, would Israel have been justified in nuking Saddam?

Ex2: Since Chris Hitchens has documented a plethora of interesting connections between Saddam and terrorists including al Qaida, how long should those conditions have been allowed to continue before the probability of there being any kind of link between Saddam and murder of American civilians could be admitted to be 1?

In other words, did we actually have to wait for more people to die after the 1st WTC bombing and subsequent flight of the perp to Iraq? How many? Some people would actually respond, in effect, infinity, since they reject military responses altogether. I think that if you categorically reject preemptive war, then it is incumbent on you to define the conditions that must be met before you react. This requires an admission that you are willing to let people die (to incur costs) before you do so. If you still think that no action is tolerable, doesn't that suggest that you are willing to bankrupt yourself to avoid the cost of war?

Cato Renasci

Paul Eremenko comments above that the abrogation of the rule against attacks unless one has been attacked or the threat is "imminent" will raise costs of self-defense for all, and is not taken into account in Judge Posner's model. This is perhaps true as far as it goes, but Mr. Eremenko does not seem to notice that on the same logic, the existence of rogue state or related non-state actors who would attack without warning similarly introduces uncertainty and raises costs. Although estimates of such costs, especially in discussions such as this, are always heuristic and imprecise, I suspect the financial and psychological costs of living with a terrorist threat at least equal and perhaps exceed (at least for the state making the cost-benefit analysis whether to engage in preventive war) any increased uncertainy and cost resulting from the preventive attack. After, a nation state actor is duty bound to take into account the costs and benefits to its society first of all and to give them primacy over the potential costs and benefits to others (except of course as they affect its own costs and benefits over time).

The other aspect of the cost benefit critiques here, as well as the more unsophisticated uses of cost benefit analysis, which I find problematic is a sense that there is a neat continuum upon which decisions can be made. (The flaw with much criticism of elementary economic theory as well) Of course, in the real world, cost-benefit functions, like demand functions, are almost certainly not convex, continuous and twice-differentialbe. There are surely discontinuities, and I take the possiblity of a major terrorist attack with weapons of mass destruction as a clear example of a catastrophic event that would represent (at least) a point discontinuity. Such discontinuities make for (and perhaps account for historically) the great difficulty in assessing after the fact the intuitive cost benefit analyses performed by decision makers with not only imperfect, but disaterously wrong information.

Since nation states, however, are in the business of making decisions under uncertainty, it seems to me we simply have to get the "best" information we can reasonably obtain and make our analysis on the basis of it. Will the decision makers sometimes get it wrong? Surely. But in dealing catastrophic events, our tolerance for the risk of being wrong (i.e. being willing to act with less certainty) surely increases as the expected costs remain high even when the probability of their occurence decreases. Is this not elementary?

[It's been too long since I read Becker and Posner's major works and was current on the literature to know if anyone has modeled this stuff as convincingly as the guys in the '70s like Werner Hildenbrand who applied measure theory to demand and equilibrium theory -- can anyone point me to the relevant literature, preferably in English as I am no longer fluent in higher mathematics?]


You dance around the question I think everybody has in mind here: should we have invaded Iraq? You almost answer it: as our information becomes sparser, the justification for war, in expectation, becomes weaker. Without good information aggregation methods, your analysis says that we can't justify a war.

It only becomes more difficult with terrorism, which is a catostrophic event. By `catastrophic', I do not mean what Gary seems to mean (causing much damage), but what the statisticians mean: you can't predict it with any reliability. There is no easy bell curve. We've been studying earthquakes for a century or two now and still don't have but the inkling of a clue as to when major earthquakes will occur; we've been seriously studying terrorism for at most a decade or two, and we have absolutely no statistical methods that begin to allow us to aggregate information into a reliable prediction of whether a terrorist act would occur.

Cato Renasci

BK -
I disagree that my analysis suggests the Iraq war was unjustified because of our failures in creating effective ("good enough?") information aggregation methods. That would be true only if one assumed that we can accurately assess the risk that our information is bad. While I agree that issue represents a systemic issue, I do not believe we have the luxury of defering decisions in media res until we have a high degree of confidence (in the theoretical sense you are speaking) in the quality of our information aggregation. After all, inaction itself, even for the (generally admirable) purpose of improving information for decision makers, represents a decision. I do not share what I take to be your bias (again in the statistical sense) in favor of inaction.
Your point about the use of catastrophic should be well taken. Although only having the been the student of students of Professor Becker, I don't think it would be a stretch to believe he was using the word in both the ordinary language meaning and the statistical sense -- surely it is possible for an event to be both.There is a tension in pieces such as these by Judge Posner and Professor Becker: they are writing indirectly about the real world, more or less in real time, using the analysis and language of economics that is inherently abstract. The interesting questions are not so much whether the abstraction is entirely correct (which no abstraction can be), but whether the abstraction is useful. At this point, I think the abstraction is useful for thinking about the problem, but I am not sure the analysis is yet sufficiently worked out to be applied convincingly to particular cases --- althougth that is an important aspect of testing their hypotheses.I'm surprised, given the criticism generally of the analogy of the Anglo-French decision to militarily oppose Hitler's remilitarization of the Rhineland (which every historian knows scared the Bejesus of the the German military, who sent the two battalions of troops in without ammunition and orders to turn tail at the slightest resistance from the Entente powers), no one has mentioned the effect of the Iraq intervention on Lybia's Col. Khadaffi. Does anyone seriously believe that Khadaffi would have renounced weapons of mass destruction and cooperated with us were it not for a sincere desire on his part to avoid the fate of Saddam Hussein?


Judge Poser begins this entire discussion by sliding this at us:

Indeed, the essence of self-defense is striking the first blow against your assailant.


How about: "the essence of self-defense is defending, deflecting, or otherwise avoiding, etc. the assailant's first blow."

Its nice for you to set up a continuium from pre-emptive to preventitive to self-defense, but you can't take the english language seriously and suggest that striking out first offensively is the essence of self defense.

Methinks you have been watching and listening to sports casters who cast about pithy aphorisms like "the best defense is a good offense."

But even these folks realize that they emphatically are not literally talking about defense.

I am 100% certain that you cannot be 100% certain that someone will attack you.

What you describe the "essence of" is preventive force. But it is most certainly offensive.


Remember that World War I was a preventive war, and that illustrates both why the Britain and France were unwilling to intervene in the Rhineland, as well as the reason why there should be overwhelming and reliable evidence before resorting to war.

In the case of WWI, the costs were vastly underestimated, while the benefits were overestimated. (Sound familiar?) So, a war that was intended to be short, cheap and definitive became one of the greatest man-made disasters of world history, with previously unimaginable costs in both money and human life. Moreover, the final outcome of the war was so inconclusive that yet another war was fought 20 years later, on roughly the same territory with roughly the same sides.

The tragedy of WWI was foremost in the minds of planners leading up to WWII, and for good reason. More than anything else, they wanted to avoid another catastrophe like WWI. This didn't blind them to the Nazi danger, but it did cause them to adjust their estimates of the costs (too high) and benefits (too low) of war with Nazi Germany. Eventually, as more information became available, and as the intentions of the Nazis became more evident, France and Britain decided to go to war. Of course, WWII caused massive destruction, but is considered to have been worth it, because of the costs, estimated retrospectively, of losing to the Nazis.

Ultimately, the greatest danger of going to war is the difficulty of correctly estimating the costs. This is an issue that, I think, is papered over too quickly in commentary here. It is also the reason why prudence dictates that war be considered as a last resort -- not because of its unique or intrinsic evil, but because it is a Pandora's box. Once it is opened, it is almost impossible to predict what will emerge.

In particular, the Bush administration's arguments for going to war with Saddam Hussein were almost entirely mistakes or lies. Even if one believes it permissible to mislead the public into supporting war against Saddam, it is vital to estimate, correctly and with caution and intelligence, the costs of that war. By now, though, it's absolutely clear that the Bush administration bungled that, too.

These issues -- of estimating costs, of intelligence and of prudence -- also hold of preventing terrorism, as well as state-to-state violence. However, the tool of war, being between states, is mostly useless against terrorist groups. Although the history of terrorism is shorter than the history of war, we already have good and bad examples of executing campaigns against terrorist groups. The Bush administration, however, seems to be unaware of these examples; they're also deliberately obscuring the real distinction between terrorist groups and nationalist insurgencies. Even if this makes good PR, it is a mistake, insofar as they seem to believe their PR.

In any case, and as a final thought, I wonder about the propriety of erasing the line between defensive and preventive war. Perhaps we should maintain it -- if only to maintain the line between preventive war and offensive war. After all, any arguments made in favor of preventive war may be applied to support offensive war as well, and the differences between them may be, at least partially, in the eyes of the beholder. I wonder how many people outside the US believe the war in Iraq was offensive, rather than preventive.


Just out of curiousity, Judge Poser tosses out .5 as a nice "round" number of probability for the likelihood of future attack.

I'm curious to know what folks think that actual "probability" was in the case of Iraq and why.


I too would like to see more writing on how to account for the costs of preventive war. The costs include, as mentioned above, the loss of perceived legitimacy. There is also a giant opportunity cost associated with going in pre-emptively. Now the US is stuck in Iraq, and has much less leverage, both diplomatic and military, against Iran and North Korea as a result of the Iraq war.

There is also the problem that the attacked region and country will view the preventive attack as illegitimate, leading to the creation of more terrorists or other groups that could attack the US in the future.

Your analysis does not seem to address the myriad, multiplying costs of preventive war as it is currently being played out.


Doesn't war have lots of external costs? Costs which the concept of imminence is, in essence, trying to roughly include into the calculation?

Or is the latest dead brown baby in a fallujah better considered a transaction cost, to be ignored?

Jason Ligon

First, as I didn't do so in the comments of the first official post, congratulations on the new blog. As an avid consumer of blogs, I have to say I'm excited about this one.

I would add to Judge Posner's comments that the benefit of attaining certainty is not to be underestimated. The probability of direct or indirect (e.g. funding or supplying terrorists) attack is something to contemplate, to be sure, but I can't shake the notion that this variable is an aggregation of:

A) What is the probability that the capability to harm us is present, and

B) What is the probability that the willingness to harm us is present?

What do we do when faced with an entity that we know has the willingness to harm us, but there is uncertainty about the capability? Eliminating the uncertainty in this case is a very large benefit of putting boots on the ground to find out.

Certainly better intelligence can help us make informed estimates of these two probabilities. I fear that many people have exaggerated notions about what intelligence is capable of, however. When dealing with an entrenched despot with a decades old internal security apparatus and a lot of geographical area in which to hide surprises, there is only so much we will know.

I am left with the image of the police officer who challenges a suspect to put his hands in the air. The suspect refuses and keeps his hands in his pockets. At some point, the impasse can be reasonably resolved by aggression.

Cato Renasci

To compare the erroneous (but almost universally so) cost-benefit analyses (intuitively) made by the Powers in July and August 1914 with Iraq is, in my view mistaken.However, I think it's not unfair to suggest that an aversion to a repeat of World War I was on the minds of the British and French governments in 1936, especially in an era when the oxford Union debated and passed (if I recal correctly) a resolution to the effect one would under no circumstances be willing to fight for King and country. And a fear of unwilling troops was not entirely unwarranted, especially in light of the French Army mutiny of 1917 and its later performance in 1940. It is useful to consider the different performances of the French and British in the Second World War in light of their histories: While WWI (especially if you consider the whole Home Rule fandango as a piece with it) represented a huge upheaval in British society, it was Englands first truly major war since the defeat of Napoleon, also an English victory. France had suffered a major defeat in 1814, and again in 1870, each of which convulsed French society from bottom to top. Although the British were war weary after WWI, they were not played out, they had the national reserves of will and determination upon which they drew in 1940, they had not lost a major war since the 18th century. The French, by contrast, had nothing left, especially if one considers revanche to have been the great unifying French idea after 1871 Having achieved it at pyrrhic cost in 1919, there was simply no idea (nearly) all Frenchmen could agree was worth fighting for.

There is a case to be made that WWI was a preventive war (the whole set of arguments about the various powers calculations of when it was to their individual advantage to fight a European war some thought inevitable), I have never found that line of reasoning quite persuasive.More likely, I would argue the series miscalulations which led to the outbreak of WWI bespeak more of a combination of failures to understand how various Powers would react to threats and the rigidity of the Powers' mobilization plans. Especially daunting was the inabilty to mobilize partially, especially on the part of the less modern Powers Austria and Russia. But we could spend entire careers debating the Kriegsschuldfrage....

The comments to this entry are closed.

Become a Fan

May 2014

Sun Mon Tue Wed Thu Fri Sat
        1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30 31