As commenters and my last post, and others, have pointed out, there’s a logical gap in my argument that, given imperfect knowledge and the recognition that we tend to overestimate our own capabilities, we should adopt a rule-based version of consequentialism which would include rules against pre-emptive or preventive wars[1]. The problem of imperfect knowledge also applies to the consequences of deciding not to start a pre-emptive war. As I’ll argue though, the symmetry is only apparent and the case for caution is strong.
I’ve addressed the underlying issue at length in a
recent paper (large PDF) trying to make sense of the well-known “precautionary principle” used in relation to poorly-understood environmental risks. I look at a range of decision procedures, from the simplest (guess what is most likely to happen and assume that will happen) through the simplistic treatment of uncertainty given by expected value models to more general approaches incorporating the recognition that, in most real world problems we will not know the probabilities of the possible outcomes and will not even have considered all the possible outcomes. Moreover, because surprises are generally unpleasant, the things that are omitted are likely to produce an overoptimistic evaluation. This leads me to state a general incompleteness hypothesis, namely
Estimates of project outcomes derived from formal models of choice under uncertainty are inherently incomplete. Incomplete estimates will generally be over-optimistic. The errors will be greater, the less well-understood is the problem in question.
The last sentence is crucial. In the context of an argument for pre-emptive war, the relevant alternative is “wait and see”. Whereas the consequences of going to war are highly unpredictable, the consequences of wait and see, over a period of, say, a few months, aren’t hard to describe. Either the putative threat will get worse, or it will fade. The cost of the wait and see approach is the possibility of having too fight later, with a less favorable balance of forces than could have been had with the pre-emptive strike. But advocates of the pre-emptive strike tend to overstate these costs and underestimate the uncertainty surrounding their preferred option.
Iraq provides a good illustration. At the time Bush and Blair decided on war, the alternative was to wait for Blix’s inspections to be completed. The reasons given for going to war in March 2003 rather than waiting until later seem absurdly trivial in retrospect. It was argued that the invasion couldn’t take place in summer and that waiting until after summer would keep forces tied up too long on standby in Kuwait. As things turned out, I’m sure Coalition forces would have far preferred summer in Kuwait to summer in Baghdad.
If Bush and Blair were actually concerned about the threat posed by Saddam, the decision to go to war in March, rather than waiting looks entirely unreasonable, except on the assumption that nothing could possibly go wrong[2]. Tim Dunlop has more on this, with specific reference to Rumsfeld’s latest observation that “you go to war with the army you have”.
More generally, the precautionary principle is not, as it might seem, symmetrical. In a situation where the consequences of one option are poorly understood, it provides grounds for avoiding, or, if possible, deferring a decision to choose that option even when a naive analysis would suggest that it should produce a better outcome. War is the paradigmatic example of an activity where “the race is not always to the swift, nor the battle to the strong, but time and chance happens to them all”. All of this leads to something close to the Powell doctrine. If war is to be an instrument of policy, it should only be used under conditions of overwhelming superiority in all phases (including occupation), for clear and feasible objectives[3], and with a clearly formulated exit strategy.
fn1. Except where the threat is so clear and imminent that standard self-defence arguments can be invoked.
fn2. An alternative, plausible in the light of the very lackadaisical attitude to weapons exhibited after the invasion is that they knew the WMD case was bogus, and needed to start the war before it collapsed altogether.
fn3. These objectives need to be justifiable in terms of the interests of the people of the world in general, and not of the national interest of one country or the personal interests of its rulers. A nation or group that pursues self-interest through military force is an enemy to all and will ultimately attract retaliation.
John Quiggin 
This all makes sense in terms of conventional warfare. But the issue now is complicated by unconventional warfare using strategic weapons.
What is the correct strategy for fighting undetectable (terrorist) and undeterrable (jihadist) enemies who are clearly prepared to use apocalyptic (thermo- and bio-) weapons?
A localised strategy of aggressive-offense (ie pre-emptive strikes) starts to look more reasonable when the enemy is so utterly alien and deadly.
OTOH a globalised strategy of passive-defence (ie rope-a-dope) may be the best way to retain coalition support, not antagonise host populations and conserve ones powder from wild goose chases.
So how do you carry out pre-emptive strikes against undetectable and undeterrable enemies?
Send yer answers on the back of a postcard to the National Security Council, Old Executive Office Building, Washington DC.
Fyodor,
I was really just trying to interpret John’s argument. But in the new post, he seems to be repudiating it, or at least saying he didn’t argue it properly. I understand the new version less than the current one, so I think I’ll just bow out of the debate. You may be completely right for all I know.
That was meant to go on the previous thread, but I guess it doesn’t matter much.
Nabakov at December 9, 2004 10:44 PM surely jests:
Round the Clock robotic terrestial, nautical and aerial surveillance and interdiction would help right now.
In the medium term, I am sure the bofins at DARPA, RAND and DSB will cook up some fiendish new gizmos that will put the fear of God into jihadists.
In the long run we should treat Fundamentalist Terrorism as virus with an epidemiological cycle. Therefore it will be best to adopt defensive measures – quarantining, immunisation and selective bug hunts – until burn-out. Draining the Swamps is not worth it.
Aging, affluencing and secularisation in the host populations should do the rest.
…and secularisation in the host populations should do the rest.
Jack might want to have a look a the Islamic Paradox by AEI scholar Reuel Marc Gerecht (download the whole book online free) who argues the exact opposite:
“But moderate Muslims are not likely the solution to bin Ladenism. Just the opposite: Those who have hated the United States most—Shiite clerics and Sunni fundamentalists—hold the keys to spreading democracy among the faithful. They, not the much-admired Muslim secularists, will probably liberate the Muslim Middle East from its age-old reflexive hostility to the West. Paradoxically, those who in their souls have felt the clash of civilizations most painfully will be our salvation from future 9/11s.”
Makes an interesting counterintuitive case:
http://www.aei.org/publications/bookID.799,filter.all/book_detail.asp
James,
Don’t sweat it. I got here eventually!
JQ,
I’m getting confused. You seem to be arguing that instead of pre-emptively attacking, states considering war should wait…until what? If a state wants to go to war because it sees a likely net gain from the outcome, and it is confident that it will win, why wait and risk the enemy getting stronger? That looks irrational to me.
You argue that,
“Estimates of project outcomes derived from formal models of choice under uncertainty are inherently incomplete. Incomplete estimates will generally be over-optimistic. The errors will be greater, the less well-understood is the problem in question.”
…yet should not the same rationale be applied to Global Warming? We would be hard-pressed to find a more complex scientific and economic problem, and yet the consensus view is that Something Needs to be Done Now.
Should not we wait for more complete understanding of the problem before restructuring our economies? What is the opportunity cost of waiting for more information and/or options? Is “do nothing” the right approach when faced with a monumental problem which we do not fully understand?
Apparently, war is a costly activity best avoided because of the scale of uncertainty surrounding the likely outcomes. But restructuring our economies must take place now despite our limited knowledge of the Global Warming process.
I’d be interested in your reconciliation of these two positions. You’re willing to advocate action in the face of uncertainty in one situation (the environment), but advocate inaction in the face of uncertainty in another (war).
I agree that the case for caution is always strong when war is involved, but I think your consequentialist argument against pre-emptive war remains unpersuasive.
Theology is the only discipline that prospers from conflating pretext and rationale.
There is little to be gained from studying the origins of the Iraq conflict from the statements made by the major belligerents. The sole utility of those statements was propagandistic. The major study that should arise from them is analysis of the modalities of willing suspension of disbelief.
There may be many wars worth analysing along the lines proposed by JQ, but the Iraq frolic isn’t one of them.
“I’m getting confused. You seem to be arguing that instead of pre-emptively attacking, states considering war should wait…until what? If a state wants to go to war because it sees a likely net gain from the outcome, and it is confident that it will win, why wait and risk the enemy getting stronger? That looks irrational to me.”
There’s a large literature on this in the context of investment which shows that it’s optimal to wait (the key terms are ‘option value’ and ‘hysteresis’). The point is that the threat might get worse, in which case you incur a small cost by waiting, or it might diminish, in which case you make a large saving by not going to war.
The paper to read is Dixit, Avinash, 1992. “Investment and Hysteresis,” Journal of Economic Perspectives
This isn’t the only relevant consideration, but illustrates one of a number of common biases, which if not countered, tend to encourage precipitate actions such as preventive war.
JQ,
The consequences of delaying military action can be deeply asymmetric. I’m not sure the optionality of investment decision-making applies, particularly in a competitive game environment. The risk isn’t that a threat might get worse. The risk is that the threat materialises and destroys your ability to exercise your option. Should you wait in that instance?
Also, if you’re the aggressor, it’s you that’s posing the threat.
‘The point is that the threat might get worse, in which case you incur a small cost by waiting, or it might diminish, in which case you make a large saving by not going to war.’
I’ll read the paper, but why does any ‘cost of waiting’ turn out to be ‘small’, whereas any ‘saving by not going to war’ turn out to be ‘large’? I can imagine that savings would turn out to be large, since wars are horribly expensive things most of the time, but need the cost really be that small? Is this a formal property of the Dixit model backed up by any empirical evidence, or just an agreeable assumption of yours?
It’s a formal property that typically characterises irreversible investment decisions, as opposed to decisions to wait. I’m assuming that going to war counts as an irreversible investment decision.
There are exceptions. For example, while it appeared plausible that Saddam was on the verge of acquiring nuclear weapons, it was reasonable to assert that waiting could result in a large discrete jump in the cost of fighting him. More generally, the kinds of arguments Posner and I are talking about here don’t really apply in the context of nuclear war. But in an ordinary context, the result follows follows from the nature of investments in all kinds.
I am puzzled by your reverse definition of status quo in the article, from a human point of view to a nature point of view. That is a crucial distinction that allows you to put forward the precautionary principle (or, actually, one of its many encarnations) as a useful framework.
I am certainly skeptical about human induced global warming, but even if you consider some of the predictions to be true, there is a lag effect on warming. Thus, if we stop emitting greenhouse gases now, it will take many years to have an effect on warming. Therefore, I would expect a ‘wait and see’ attitude more rational than an abrupt intervention.
Hasn’t a logic trap been built into this system:
time n: “the threat might get worse, in which case you incur a small cost by waiting”
[the threat does get worse]
time (n + 1): “the threat might get worse, in which case you incur a small cost by waiting”
[the threat does get worse]
time (n + 2): “the threat might get worse, in which case you incur a small cost by waiting”
[and so on]
Doesn’t this imply that the moment when the “precautionary principle” ceases to operate is asymptotically approached but never reached?
No, Katz, your argument doesn’t work. If the expected benefits of investment are large enough, they outweigh the gain from waiting. If you want to follow the debate up, it’s well worth taking a look at Dixit, Avinash (1992), “Investment and Hysteresis”, Journal of Economic Perspectives 6(1, Winter).
Much of this is a restatement, with a more complex model of points raised in the 1970s by Arrow, Fisher Hanemann and Henry.
Looking here, it appears that Fisher has claimed, correctly in my view, the two are equivalent, but that others have denied this.
“At the time Bush and Blair decided on war, the alternative was to wait for Blix’s inspections to be completed.”
Don’t forget the waiting for Saddam Hussein and then Qusay Hussein to die of old age. It probably wouldn’t have taken more than…maybe 30 more years. Or 40.
Conservationists and conservatives
Don Arthur had an interesting response to my pieces on the precautionary principle and wars of choice1. Don correctly observes that this kind of argument can be used in opposition to reform, and is therefore inherently conservative. He mentions, as…