2 potential aggressor, including its theory of deterrence (taking intoaccount what it values and why). In the process, as will be argued,history strongly suggests that aggressor motivations are varied andcomplex, and as often grounded in a desperate sense of a need toact as they are the product of aggressive opportunism.2 Deter- rence turns out to be about much more than merely threateninga potential adversary: It demands the nuanced shaping of percep- tions so that an adversary sees the alternatives to aggression as moreattractive than war. Defnitions and Types Deterrenceis the practice of discouraging or restraining someone— in world politics, usually a nation-state—from taking unwantedactions, such as an armed attack. It involves an efort tostoporprevent an action, as opposed to the closely related but distinctconcept of “compellence,” which is an efort toforcean actor to dosomething. Denial Versus Punishment Te classic literature distinguishes between two fundamentalapproaches to deterrence.Deterrence by denialstrategies seek todeter an action by making it infeasible or unlikely to succeed, thusdenying a potential aggressor confdence in attaining its objec- tives—deploying sufcient local military forces to defeat an inva- sion, for example.3 At their extreme, these strategies can confronta potential aggressor with the risk of catastrophic loss. Deterrenceby denial represents, in efect, simply the application of an inten- tion and efort to defend some commitment. A capability to denyamounts to a capability to defend; “deterrence and defense areanalytically distinct but thoroughly interrelated in practice.”4 Temost common way of measuring the health of a deterrence threatgrounded in denial capabilities is the immediate balance of forcesin the contested territory—but, as will be explained, the local bal- ance of forces is not the only, or even always the most important,factor. Deterrence by denial should not be equated with militarybalances alone. Deterrence by punishment , on the other hand, threatens severepenalties, such as nuclear escalation or severe economic sanctions,if an attack occurs. Tese penalties are connected to the local fghtand the wider world. Te focus of deterrence by punishment is notthe direct defense of the contested commitment but rather threatsof wider punishment that would raise the cost of an attack. Most classic studies suggest that denial strategies are inherentlymore reliable than punishment strategies.5 Steps taken to deny, suchas placing signifcant military capabilities directly in the path of anaggressor, speak loudly and clearly. An aggressor might doubt, onthe other hand, a defender’s willingness to impose punishments.6An aggressor might also convince itself that the defender willhesitate to follow through on threats to punish because of attendantrisks, such as further escalation, that the deterring state may notAn aggressor might also convince itself that the defender will hesitate to follow through onthreats to punish because of the attendant risks that the deterring state may not be willing torun once the moment arrives. 3 be willing to run once the moment arrives.7 As Tomas Schellingnoted, there are threats that a state would rather not fulfll, andweakness in deterrence can emerge when an aggressor believes thedefender will ultimately prove unwilling to carry out its threats.8 Direct Versus Extended Deterrence can be used in two sets of circumstances.Direct deter- renceconsists of eforts by a state to prevent attacks on its ownterritory—in the U.S. case, within the territorial boundaries ofthe United States itself.Extended deterrenceinvolves discourag- ing attacks on third parties, such as allies or partners. During theCold War, direct deterrence involved discouraging a Soviet nuclearattack on U.S, territory; extended deterrence involved preventing aSoviet conventional attack on North Atlantic Treaty Organization(NATO) members.9 For obvious reasons, extended deterrence is more challengingthan direct deterrence. Tis is partly true for military operationalreasons: It is more difcult to deny an attack far from home, amission that demands the projection of military force sometimesthousands of miles away and often much closer to the territory ofthe aggressor state. However, it is also true for reasons of credibil- ity. An aggressor can almost always be certain a state will fght todefend itself, but it may doubt that a defender will fulfll a pledgeto defend a third party. During the Cold War, for example, therewere constant debates about the credibility of the U.S. promise to“sacrifce New York for Paris.”Reinforcing extended deterrence involves taking steps to con- vince a potential aggressor that the distant defender will defnitelyrespond to an attack, or at least as promptly as it can in accordancewith national laws. Such steps include actions like stationing signif- icant numbers of troops from the deterring state on the territory ofthe threatened nation, as the United States has done in many cases.Te defender seeks to create the perception that it has, in efect, nochoice but to respond if its ally is attacked. Yet this is a demanding standard to meet, in part because astate will seldom commit to anything like an automatic response ifvital national interests are not at stake—and often, even if they are.Te most famous cases of extended deterrence failure involving theUnited States—such as Korea in 1950 and Iraq-Kuwait in 1990— can be partly traced to the fact that the United States was unwill- ing to demonstrate automaticity of response before the fact. Eventhe most powerful treaty commitments generally contain somedegree of leeway. Article 5 of the North Atlantic Treaty, which isarguably the strongest U.S. commitment of extended deterrence,does not oblige parties to take an automatic response to aggres- sion against any other ally. It calls on parties to take “forthwith,individually and in concert with the other Parties, such action as itdeems necessary, including the use of armed force, to restore andmaintain the security of the North Atlantic area.”10 Tis languagerefected a compromise between the United States’ Europeanallies, which sought as close as possible to an automatic response inthe event of aggression, and the U.S. Congress, which wanted topreserve its war powers. Similar conditions can be found in all U.S.mutual security treaties. Te United States has sometimes hesitated to make less- ambiguous deterrent threats, such as in the cases of Korea and Iraq,because of another complication in extended deterrence (and deter- rence threats of all kinds): Such threats can be very costly to make.Tis is partly true because of the implied commitment involved— once the United States has threatened to respond to a certain sort4 of attack, it must then plan and prepare to do so. Much of thecurrent U.S. defense budget is devoted to building the capacity andcapabilities necessary to engage in the large-scale contingencies thatrepresent the U.S. global deterrence posture. But threats can also becostly in diplomatic terms, generating deeper tensions with rivalswho may or may not have been intending to attack. Defenders, therefore, are constantly engaged in a tenuousbalancing act. Tey are trying to gauge the national interests theyhave at stake in a potential contingency, the costs and risks of beingvery explicit about their response, and the dangers of aggression ifthey do not make such explicit threats. Such complex dynamics areapparent in the U.S. and NATO eforts to warn Russia of aggres- sion in the Baltic states today. General Versus Immediate Finally, the theoretical literature distinguishes between two over- lapping time periods in which deterrence policies can be employed.General deterrenceis the ongoing, persistent efort to preventunwanted actions over the long term and in noncrisis situations.Immediate deterrencerepresents more short-term, urgent attemptsto prevent a specifc, imminent attack, m