Barack Obama’s campaign promise to close Guantánamo hardly put him out on a limb. Republican Senator John McCain, Secretary of State Condoleezza Rice, President George W. Bush, and Defense Secretary Robert Gates have all said that they would like to close it. The facility is now a more potent symbol of the United States worldwide than the Statue of Liberty, a daily reminder of the lawless approach the outgoing administration has taken toward the so-called “war on terror.” With broad support, Obama will likely close Guantánamo next year.

But closing Guantánamo will raise almost as many problems as it solves. Where, for example, will the new administration put the 250 or so men still detained there? No American city is eager to have accused al Qaeda terrorists in its backyard. And outsourcing the problem is not an option. Most of the detainees’ native countries will not take them back; cannot assure us that the men won’t return to battle; or may torture the detainees, thereby precluding us from repatriating them.

The difficult question of relocation, however, pales in comparison to the long-term problem of what to do with the remaining detainees. The Bush administration claims that about eighty could be criminally tried. If we cannot try the remaining 170, must we release them?

Many human rights advocates say yes. In their view, absent a war with another nation, neither the laws of war nor American laws authorize extended preventive detention. If a Guantánamo detainee is guilty of a crime, he can be tried and, if convicted, sentenced to a period of incarceration. Those who cannot be tried or convicted should be released. These advocates argue that any form of preventive detention is a dangerous departure from the existing paradigm of crime and punishment, and ought not be tolerated.

Others, such as my Georgetown University colleague Neal Katyal and Harvard Law Professor Jack Goldsmith, recommend that Congress grant the president authority to detain indefinitely anyone who can be shown to be a “suspected terrorist.” They argue that the threat of terrorism warrants a sweeping new preventive detention authority and favor the creation of a specialized “national security court” to administer it.

In my view, Obama should adopt neither of these alternatives. The try-or-release approach leaves us without sufficient protection in modern-day military conflicts. At the same time, authorizing preventive detention of suspected terrorists, as Katyal and Goldsmith advocate, would be both unconstitutional and unwise. There is no justification for bypassing the criminal justice system simply because an individual is suspected of terrorism rather than drug dealing, rape, or murder. If we were to create such an exception for terrorists, what would stop its extension to other serious crimes?

There is, however, a third alternative—one that allows the United States sufficient authority to protect itself from al Qaeda fighters while avoiding the creation of an exception that threatens to swallow the rule of the criminal process. Congress should follow the example of traditional wars and give the administration the option to detain—without criminal trial—those engaged in hostilities with us for the duration of the military conflict with al Qaeda and the Taliban. Detainees should be afforded punctilious procedure to ensure that we are detaining only those who fought for al Qaeda or the Taliban and pose an ongoing threat.

Detention without trial must be a carefully circumscribed exception during peacetime, but it has long been recognized as an appropriate and necessary means of dealing with enemy fighters during wartime. If the United States could lock up German soldiers during World War II without trying them criminally, why should it not have the same option for al Qaeda fighters? The fact that the armed conflict with al Qaeda is not a war between nations ought not disable the government from holding its enemies preventively while the conflict goes on.

So the problem with Guantánamo Bay, as I have long argued, is not the detention of enemy combatants. The problem is that the Bush administration has denied fair hearings, resulting in the detention of many who were not enemy fighters; it has defined the category of “combatants” so as to sweep in grandmothers who donated unknowingly to an organization affiliated with al Qaeda; it has authorized cruelty and torture as interrogation methods; it has asserted the right to hold prisoners for the duration of the “global war on terror” (that is, forever); and, most fundamentally, it has argued that Guantánamo is a law-free zone. Guantánamo is a black mark because of this resistance to law and refusal to recognize the basic human dignity of the detainees. If we are to fix the problem, we need not abandon military detention, but we must subject it to the rule of law.

• • •

There are many reasons to be skeptical of preventive detention. They begin with experience—we have not done preventive detention well. Consider the three most aggressive domestic preventive detention campaigns in U.S. history (not counting the Civil War): the Palmer Raids of 1919-20, the Japanese internment of World War II, and the widespread preventive detention of foreign nationals in the United States after 9/11. In each instance, the United States responded to a violent attack by rounding up the usual suspects, based not on concrete evidence of involvement in the violence, but on the basis of much broader “profiles”—Communist Party membership, Japanese ancestry, or Arab or Muslim identity. None of those locked up in the Palmer Raids was found to have engaged in the bombings that prompted the raids. None of the Japanese or Japanese-American internees was a spy or saboteur. And none of the more than 5,000 foreign nationals jailed in anti-terrorism preventive detention measures in the first two years after 9/11 stands convicted of a terrorist offense.

But the problem with preventive detention is not just that we have managed it poorly. For three reasons, it is an inherently dangerous enterprise. First, no one can predict the future. Preventive detention turns on predictions of future harms that cannot ultimately be proven or disproven. In the absence of an ability to predict, we often resort to inaccurate stereotypes and prejudices as proxies for dangerousness. Preventing harm is a legitimate social goal, of course, but there are many ways to do so short of detention, such as securing borders, enhancing intelligence-gathering, safeguarding nuclear stockpiles, and engaging in smarter foreign policy. Locking up human beings is one of the most extreme preventive measures a state can undertake; it should be reserved for situations in which it is truly necessary, but uncertainty about the future makes necessity virtually impossible to establish.

Second, and relatedly, the risk of error—in particular, the error of unnecessarily detaining innocent people—is high. The detention process will undoubtedly be skewed by the fact that some kinds of errors are highly visible, while others are entirely invisible. When a judge erroneously releases an individual who poses a real danger of future harm, and the individual goes on to commit that harm, the error will be emblazoned across the front pages, there for all to see. When, by contrast, a judge detains an individual who in fact would not have committed any wrong had he been released, that error is invisible—and, indeed, unknowable. (How can one prove what someone would not have done had he or she been free?) Thus, human nature suggests that judges overseeing preventive detention will err on the side of custody over liberty.

Third, preventive detention is inconsistent with basic notions of human autonomy. We generally presume that individuals have a choice to conform their conduct to the law. For this reason, liberal societies do not criminalize thoughts or intentions, but actions. Dangerous activity can usually be criminally proscribed, and we ought to trust, absent some very strong showing, that individuals will obey those proscriptions. To imprison a human being on the claim that he will take dangerous and illegal action if we do not is to deny his autonomy.

Any system of preventive detention inevitably poses these problems. We can mitigate them, but they cannot be eliminated. Accordingly, any consideration of preventive detention should begin with a strong constitutional presumption that our society deals with dangerous people through criminal prosecution and punishment, not preventive detention. We should depart from the criminal justice model only where the criminal process cannot adequately address a particularly serious danger.

And, indeed, U.S. law has long recognized the propriety of preventive detention where the criminal justice system is demonstrably inadequate. We permit civil commitment of persons who, because of a mental disability, literally cannot be held responsible by the criminal justice system—they lack the requisite intent to be held culpable. We authorize detention without bail of persons facing criminal trial or deportation when they pose a danger to the community or a risk of flight. We cannot instantaneously adjudicate criminal liability or immigration status, and so while the system proceeds, preventive detention is allowed. Quarantines similarly fit this model. We cannot make it a crime to have a disease, so quarantines rely on preventive detention to protect the community from a danger that the criminal justice system cannot address.

Absent similar showings that the criminal process is inadequate, preventive detention is an unconstitutional infringement on liberty. Thus, there is no justification for creating a preventive detention regime aimed at suspected terrorists, as Katyal and Goldsmith advocate. A recent report by former prosecutors Richard Zabel and James Benjamin, Jr., prepared for Human Rights First, analyzed 107 terrorism cases prosecuted in American courts since the 9/11 attacks. The report shows that the criminal justice system has proven capable of incapacitating, trying, and convicting terrorists before and after 9/11. Terrorism is a serious crime, but just like other serious crimes, it can and should be addressed through the criminal justice system.

• • •

We need to find a way to address serious and legitimate security concerns without the overkill of a general system of preventive detention for suspected terrorists. An alternative approach, sensitive to both security and liberty, would permit preventive detention, but only for detainees identified as fighters in an ongoing military conflict.

Preventive detention has long been accepted in wartime, precisely because the criminal justice system cannot address the problem of incapacitating enemy soldiers. In a traditional war, enemy soldiers are generally “privileged” to fight, and therefore we cannot make it a crime for the soldier to fight for the other side. In addition, we cannot presume that enemy soldiers will conform their actions to our laws by avoiding combat if we release them, because they are compelled by their own countries’ laws to fight. Finally, problems of proof regarding battlefield captures and the need to maintain military secrets during an armed conflict mean that criminal prosecution will often be practically foreclosed even where it is a legal possibility. No one disputes, then, that a nation fighting a traditional war against another nation has the right to capture and detain enemy soldiers for as long as the conflict lasts.

But the conflict with al Qaeda is not a traditional armed conflict. Al Qaeda is not a state, has not signed the Geneva Conventions, is difficult to identify, and targets civilians. Nevertheless, we are in an armed conflict with al Qaeda. The “global war on terror” is an ill-conceived metaphor or slogan, but the military conflict with al Qaeda is real. Al Qaeda declared war on the United States, and has attacked it both at home and abroad. The attacks of 9/11 were of such a scale that both NATO and the United Nations General Assembly recognized that a military response in self-defense was justified. Approximately 120 nations signed on to the United States’ invasion of Afghanistan after the Taliban refused to turn over Osama bin Laden, al Qaeda’s leader. The conflict continues to this day and, if anything, appears to be growing worse.

Unlike an opponent in a traditional armed conflict, however, al Qaeda is an “unprivileged belligerent”; it has no right to wage war against the United States. Its actions can be, and for the most part have been, criminalized. Because the criminal option remains available, the case for preventive detention here is less clear-cut.

Still, there are many reasons why the criminal process is insufficient to incapacitate the enemy in an armed conflict—even an enemy such as al Qaeda, which is subject to criminal prosecution under the laws of war. These include the difficulty of collecting and maintaining evidence in wartime settings, the heightened need for secrecy in an ongoing military conflict, and the possibility that enemies might use the criminal process to pass information to their compatriots.

In addition, the burden of proof in criminal cases, including war crimes cases, is very high—the government must prove culpability beyond a reasonable doubt. Suppose that the government has “clear and convincing evidence” that an individual was captured while actively engaged in armed conflict on behalf of al Qaeda, and that, furthermore, the individual has boasted that he would return to the struggle if released. Now suppose also that the government is unable to convince a jury—civilian or military—that the individual is guilty beyond a reasonable doubt of a specific crime. Must he be released? An Italian soldier who prevailed in a war crimes trial during World War II would not be entitled to release upon acquittal but only upon the cessation of hostilities. Why should an unprivileged belligerent fighting for an entity that has no right to fight receive better treatment than an Italian soldier fighting for Italy during World War II?

Put simply, the fact that al Qaeda is engaged in criminal warfare should not restrict the United States’ options in defending itself. We certainly have the right to try al Qaeda fighters for war crimes, and we also have the right to try them for ordinary crimes. But should we be required to try them in either forum, particularly while the conflict is ongoing? In a traditional international armed conflict, the fact that a given detainee may be tried as a war criminal does not mean that he must either be tried or released. War crimes trials typically occur at the conclusion of a war because a nation at war has a strong interest in focusing its resources on the conflict itself and in not revealing what it knows about the enemy. In the meantime, the suspected war criminal may be held as a combatant for the duration of the conflict, whether or not he is ever criminally tried.

Thus, traditional concepts of justice during wartime do allow for preventive detention to forestall a combatant’s return to the field of battle. Suspicion of terrorism is no basis on which to hold detainees preventively, but active engagement in an ongoing armed conflict is.

• • •

The most important issue in devising an acceptable, limited preventive detention regime is the definition of who may be detained. We need to distinguish those who are actively engaged as belligerents, albeit in an unconventional war, from those who are not.

In Hamdi v. Rumsfeld, the Supreme Court ruled that the executive could hold persons captured on the battlefield in Afghanistan fighting for the Taliban or al Qaeda against the United States. This is surely the core case, and few would deny that such fighters are subject to military detention. But how far should the detention power extend beyond that? What about people captured far from the battlefield? What about members of the Taliban who have never fought against the United States? What about those who sympathize with al Qaeda, and may even be inspired by it to engage in terrorism, but remain independent actors? What about someone who provides financial support to al Qaeda or the Taliban, but is not a member of either? What about a doctor who has provided medical attention to a Taliban fighter?

The Bush administration took an extraordinarily sweeping view of who may be detained as an enemy combatant. It defined the category as containing not only members of al Qaeda or the Taliban, but also those “associated” with these groups; those who have merely “supported” these groups; and those who are members, associates, or supporters of other groups “affiliated” with al Qaeda or the Taliban.

This goes too far. If one analogizes to World War II, for example, we were entitled to detain as enemy combatants anyone who fought for the German armed forces, but not anyone who paid taxes in Germany, treated a German soldier in a hospital, or was a member of a “Friends of Germany” association.

In a case the Supreme Court has agreed to hear, the U.S. Court of Appeals for the Fourth Circuit recently addressed the question of who may be detained as an enemy combatant. While residing in the United States, Ali Saleh Kahlah al-Marri, a citizen of Qatar, was transferred from civilian to military custody shortly before he was to go on trial for criminal charges relating to identity fraud and lying to FBI agents. Al-Marri is the only enemy combatant currently in military custody in the United States. The United States alleged that al-Marri trained in an al Qaeda training camp, worked closely with and took orders from the al Qaeda leadership, and came to the United States as an al Qaeda agent for the purpose of engaging in terrorist activities in the United States. In a splintered opinion, the court held that if the allegations against him were true, he could be detained as an enemy combatant. But it also found that he had not been afforded a fair hearing.

Four dissenting judges maintained that only those captured on a foreign battlefield or foreign soil could be detained as enemy combatants. This seems too restrictive. If an enemy fighter is captured outside the field of battle, but the capturing nation has reason to believe that if left free the fighter would resume hostilities against it, why should it be compelled to release him? In World War II, the Supreme Court upheld the detention and military trial of several members of the German armed forces, including an American citizen, who were captured in various U.S. cities, far from any battlefield. Moreover, where the enemy does not recognize any limits on whom it may target or where it may attack, restricting military detention to those found on traditional battlefields would unreasonably hamstring U.S. defenses.

The judges in the majority agreed that the dissenters’ definition was too narrow, but could not agree on an alternative. Judge J. Harvey Wilkinson’s opinion is the most convincing. He would require the government to establish that an individual is (1) a member of (2) an organization against which Congress has authorized the use of military force (3) who “knowingly plans or engages in conduct that harms or aims to harm persons or property for the purpose of furthering the military goals of an enemy nation or organization.” The first two criteria, Wilkinson explains, establish whether the individual is an “enemy”—a term that encompasses only those formally members of an entity against which Congress has authorized the use of military force. The third criterion determines whether the individual is a “combatant,” and serves to distinguish “mere members” from those actually engaged in hostilities on behalf of the enemy.

These criteria would be an excellent guide for legislation; what is less clear, and what the U.S. Supreme Court will now consider, is where the authority to detain such individuals comes from in the absence of express congressional authorization.

The Israeli Supreme Court has also recently addressed the issue of who may be detained in an armed conflict with a terrorist organization—in this case, Hezbollah. It authorized detention where the government proves that an individual either (1) took a non-negligible part in hostilities against Israel, or (2) was a member of an organization engaged in such hostilities and “made a contribution to the cycle of hostilities in its broad sense.”

The approaches of Judge Wilkinson and the Israeli Supreme Court differ in important ways, but they share certain features that should be at the core of any conception of enemy combatants. First, both treat detention as justified only by a military conflict, a requirement that significantly checks the power to detain preventively. The United States has suffered numerous terrorist attacks over the course of its history but has authorized the use of military force in response only once. Furthermore, armed conflicts, even where they are extended, come to an end someday. Thus, a military detention power is a temporary authority with a defined end point, even if one cannot predict precisely when it will come. A preventive detention statute for suspected terrorists, by contrast, would be a permanent feature of the law.

Second, neither Judge Wilkinson nor the Israeli Supreme Court would permit detention of mere supporters of the enemy organization, much less supporters of affiliated groups. And, equally significantly, neither would permit detention based on membership alone. Why? Because unlike a military force, a terrorist organization is a political organization, and one cannot presume that all of its members are involved in hostilities.

Finally, neither the Israeli Supreme Court nor Judge Wilkinson would restrict military detention to battlefield captures, for the reasons I noted above.

In short, were Congress to so provide, a limited group of individuals could constitutionally be subject to detention in the ongoing military conflict with al Qaeda and the Taliban. That category of enemy combatants should be confined to (1) persons involved in actual hostilities with the United States on the part of al Qaeda or the Taliban; or (2) members of al Qaeda or the Taliban who can be shown, by their activities or their positions in the organization, to have played a direct role in furthering its military ends through training fighters or planning, directing, or engaging in hostile military activities.

• • •

Defining the category of enemy combatants is difficult and important. Almost as difficult and important is ascertaining whether an individual is in fact a combatant. What kind of process should we follow in determining who is legitimately detained?

The Bush administration initially insisted that the Guantánamo detainees were entitled to no process whatsoever. After the Supreme Court held in Hamdi that citizens were entitled to due process, the administration hastily created “Combatant Status Review Tribunals.” In these CSRTs, detainees were not allowed the assistance of a lawyer, the tribunals heard no live testimony, and much of the evidence was confidential and hidden from the detainee, making a meaningful rebuttal impossible. Moreover, the hearing officers were subordinates of commanders who had already determined—without a hearing—that the detainees were enemy combatants.

The CSRT process is insufficient, particularly given the lengths of the detentions and the challenge of distinguishing genuine enemy combatants from innocents turned in by bounty hunters. According to the Hamdi decision, due process requires that the detainee be afforded notice of the factual basis for his detention, a meaningful opportunity to rebut that argument, and a neutral decision-maker. Even this, however, is only a starting point, which leaves at least four basic questions unresolved: Do the same due process rights apply to foreign nationals? What is the burden of proof? Are detainees entitled to lawyers? And how should confidential information be handled?

First, there ought be no double standards in the treatment of citizens and foreign nationals. The due process inquiry balances the individual’s interest in liberty against the government’s interest in security. On both sides of the scale, these interests are the same whether the suspected enemy is a foreign national or a U.S. citizen. Therefore, the procedural guarantees ought not differ.

Second, the government should be required to establish the need for detention by “clear and convincing evidence”—a standard that falls between the standards for criminal and civil cases. Periodic reviews of the detainee’s legal status should be required, and, as the Israeli Supreme Court ruled, the longer a detention continues, the stronger the government’s showing should have to be.

Military detention is always indefinite because one cannot know whether the war will last months, years, or decades, but at least we know what the end of a traditional war looks like. It is difficult to predict what form the end of the conflict with al Qaeda will take. This difference ought not eliminate the possibility of preventive detention altogether, but it should require more careful and regular procedures for assessing and reassessing whether continued detention is necessary.

Third, detainees must be afforded a truly independent assessment of the facts, as well as the assistance of a lawyer. Leaving detention decisions to military subordinates is deeply problematic, as two recent federal court decisions ordering the release of Guantánamo detainees for lack of sufficient evidence vividly demonstrate. In both cases, CSRTs rubberstamped the military’s detention decisions, but when independent judges reviewed the evidence in an adversarial setting with legal representation for the detainees, they found it wanting.

Finally, one of the most difficult procedural issues lies in reconciling the individual’s right to a fair hearing with the state’s interest in maintaining secrecy. While the military often has a legitimate interest in maintaining the confidentiality of relevant information in an ongoing conflict, it should be allowed to use confidential information only if (1) it has exhausted all options that might protect both its interest and the interest of the detainee; and (2) its use does not defeat the individual’s meaningful opportunity to respond. There are numerous options available to reduce the unfairness of secret evidence, including declassification review, the use of unclassified summaries, and provision of security clearances for defense attorneys so they may have access to the classified evidence. The bottom line must be that indefinite detention cannot be imposed unless the detainee has a meaningful chance to defend himself. In the case of an irreconcilable conflict, liberty must prevail.

Congress has thus far left the regulation of enemy combatant detentions to executive innovation. Considering the hazards involved—both for the detainees, who have already spent years in detention, and for the United States, whose reputation has been severely damaged worldwide by its failure to accord the detainees a fair process—a statute setting forth carefully crafted and fair rules for enemy combatant hearings is critical. And as court decisions have shown, if Congress does not act, the courts will.

• • •

Some will object that establishing such a preventive detention regime may open the door to the use of preventive detention against organized crime, drug gangs, and terrorists generally. This slippery slope would be a genuine concern were we to adopt a statute focused on suspected terrorists, because such a statute would not be limited to wartime. There is little doubt that there is an ongoing military conflict in Afghanistan. The same has never been true with respect to drugs, organized crime, or indeed, most acts of terrorism. Even if we occasionally deploy the military to interdict drug dealers abroad, Congress has never “declared war” against drug dealers in the non-metaphorical way it has declared war against al Qaeda. The situations in which war will be a legitimate response are likely to be exceptional, and thus the requirement of an ongoing military conflict should reserve preventive detention to where it is truly necessary.

Any preventive detention regime undeniably presents substantial risks that individuals will be detained unnecessarily. One might reasonably conclude that we ought not go down this path in the first place. But then one would need to show why all the forms of preventive detention that the United States and other liberal democracies already tolerate are not equally illegitimate. The real question is why military detention of enemy fighters in an ongoing armed conflict ought not be as appropriate a basis for preventive detention as civil commitment of the dangerous and mentally incompetent, or a quarantine of those who may spread disease. No society of which I am aware rejects all preventive detention. We would do best to control the risk of expanding this inherently dangerous tool by adopting a regime justified only as military detention, a concept with narrow and well-established parameters.

There is no doubt that Guantánamo needs to be closed. But closing Guantánamo will do little to resurrect our reputation worldwide unless we simultaneously institute an acceptable way of dealing with the detainees after that lone prison is shuttered. Releasing all who cannot be convicted criminally is not a realistic option as long as the war is ongoing and they pose a real threat. Creating an entirely new concept of detaining suspected terrorists invites a slippery slope and fails to satisfy the threshold requirement of showing that the criminal justice system is not up to the task. Resting our policy on accepted tools of military conflict seems far more likely to achieve the security—and the legitimacy—we need.