Scholars at the Cato Institute have argued that the use of military force can be justified when vital American interests are at stake, public support for engagement exists, we are likely to prevail, and there are no better options. Of course, each of those conditions raises numerous additional questions, and reasonable people can disagree about specific cases. After 9/11, Cato scholars backed narrowly defined military involvement in Afghanistan. But we have rejected adventurism that merely seeks national greatness, promotes nation-building, encourages military responses to imagined crises, transfers wealth to corrupt dictators, or unduly depends on the efficacy of government.

We endorse self-defense: the use of retaliatory force against domestic and foreign predators who have initiated force against us. We recognize, however, that the application of that principle can be every bit as difficult as defining, in the context of military intervention, vital interests, public support, the probability of victory, and the costs and benefits of alternative responses.

Resources are always limited. Efforts to counteract an identified danger will necessarily diminish private investment or resources available for other government pursuits. Accordingly, we must not rely on precautionary logic, the notion that government must take preventive action whenever an activity poses a possible risk. Instead, we have to quantify not only the prospective risk but also the probability of its occurrence. Yes, a low-probability chance of a catastrophic event can justify extensive defensive actions, including military force. But pretending that a catastrophe is certain can lead to dangerous and wasteful preventive measures.

Using retaliatory force is particularly troublesome when punishing “endangerment” —that is, conduct that might violate the rights of others but hasn’t yet done so. How much increased risk do I have to endure before your potentially (but not yet) actually harmful behavior is restrained? Surely government can impose safety requirements on nuclear power plants without waiting for a fatal accident. But should government be able to compel automobile manufacturers to install expensive antipollution equipment? How about retaliation againstnuisances, such as loud noises or bright lights? When do these sorts of behaviors violate the rights of others?

In the foreign policy arena, at what point, if at all, should the United States intervene militarily? Suppose we were to establish that Iran or North Korea was developing an offensive nuclear capability. Bombing those countries might stop their aggressive behavior, but deterrence must be evaluated at the margin. How much added deterrence do we get by using our military versus alternative means of persuasion that might cost less, sacrifice fewer (or perhaps zero) American lives, and garner more support from our allies?

Cato’s critics insist that predators must be swiftly and forcibly stopped. Police encountering a bank robber, for example, don’t negotiate peaceful alternatives whereby the criminal can keep half his loot if he promises not to rob again. But what if the property stolen belongs legitimately to the robber; or the bank is owned by Kim Jong Un; or the thief has hostages; or he possesses a bomb or explosive vest? Meeting the robber unhesitatingly with retaliatory force might be the worst way to secure our rights.

Nearly every act imposes some risk on others — driving, carrying a gun, lighting a campfire, developing a nuclear weapon. Which of those acts represents the initiation of force that justifies retaliatory action? When considering borderline cases such as endangerment, sensible observers understand the need to apply pragmatic, utilitarian cost-benefit measures — for example, the magnitude and probability of the harm that might occur, the likelihood that government action will remedy the problem, the availability of alternative remedies, and the cost of intervening.

The complexities of foreign policy and military engagement cannot be shoehorned into a Procrustean rule endorsing retaliatory force or precautionary intervention. Foreign policy can entail collateral damage and grave unintended consequences. Therefore, force is rarely if ever the first and best resort. Nor should we assume prescience on the part of our elected officials — the same officials who gave us Vietnam and Iraq. Beware the fatal conceit that government planners can gather and process dispersed and imperfect knowledge to reshape the world according to their wishes.

The purpose of a principle is to allow us to characterize the vast majority of cases as falling on one side of the line or the other. But there will always be hard cases — ambiguity at the margin — that we will have to tolerate and resolve pragmatically case by case. Sometimes the luxury of clarity is accorded only to those who refuse to see the fog.