Amid the new round of headlines about the Bush administration’s use of extremely harsh methods–some say torture–to interrogate suspected terrorists, the most important question is getting the least attention.

That question is how we should revise our laws to govern future interrogations, especially those of newly captured terrorism suspects who seem especially likely to have potentially lifesaving information.

Debate currently focuses on whether evidence obtained through highly coercive methods should be used in military commission trials of those accused of being Qaeda leaders,

and whether President Bush and his top national security aides should be investigated for war crimes for approving those methods.

My answer is no to both questions. But looking forward, the key to a decent, effective, democratically legitimate interrogation policy is for policy makers and citizens to think hard about an agonizing choice.

What would we want done the next time the CIA catches someone such as Khalid Shaikh Mohammed (KSM, in official shorthand)? There is no good answer. But we should be able to improve on the disastrous legacy of Bush, whose presidency brought us the horrors of Abu Ghraib and the "renditions" of at least two wrongly suspected men to be tortured in Syria and Afghanistan.

The March 2003 capture of KSM in Pakistan was as close to the hypothetical "ticking bomb" situation as we have come in the real world. Although the CIA did not know of any imminent threat, it did know that KSM was the architect of the 9/11 attacks and was Al Qaeda’s chief of operations. As such, he probably knew more than anyone else alive about any planned attacks and where to find other key terrorists.

The CIA thus had reason to believe that unlocking the secrets in KSM’s head might save dozens or hundreds of lives–and perhaps many, many more, in the unlikely but then-conceivable event that Al Qaeda was preparing a nuclear or biological attack on a major American city.

The CIA also had reason to believe that the only chance of extracting important information from KSM was to break his resistance by using highly coercive, ugly methods amounting to torture (a crime) or near-torture (deemed a crime by many experts). This tough, smart, committed jihadist was not about to betray his cohorts to his hated enemies if interrogators stuck to the kid-glove interrogation rules demanded by human-rights groups and, recently, by most congressional Democrats.

Under such rules, the CIA would have had to abandon any serious effort to save lives that might depend on learning KSM’s secrets. Some philosophers make a strong moral case for such abstinence. But no president who takes seriously his or her responsibility to protect the American people–not a Hillary Rodham Clinton, not a John McCain, and not (I hope) a Barack Obama–would want to be bound by such rules.

Would a wise president have gone as far as the Bush administration authorized the CIA to go–or as far as it went–with KSM? The answer might depend on the likelihood that lives could be saved, which will always be exceedingly difficult for any president and impossible for outsiders to gauge.

To break KSM and two other suspected Qaeda leaders, the administration resorted to waterboarding–simulated drowning–which is deemed by many experts to be torture but was then defended by administration attorneys as legal, if limited in duration. KSM did not easily break, however. So "a variety of tough interrogation tactics were used about 100 times over two weeks," The New York Times reported last October 4. "Agency officials then ordered a halt, fearing the combined assault might have amounted to illegal torture."

Did all this brutality work? Did it save lives?

Bush and others, including Director of National Intelligence Mike McConnell, have so claimed. KSM gave up "information that helped us stop another planned attack on the United States," described "other plots to kill innocent Americans," helped U.S. agents find other terrorist leaders, and led to the breakup of a Qaeda cell that was "developing anthrax for attacks against the United States," Bush said in September 2006.

Of course, Bush has little credibility on such matters. McConnell has more. But FBI officials and even some former CIA officials have disputed CIA claims that the agency has had great success with Qaeda captives by using coercion. So we can’t be entirely confident that brutalizing KSM saved lives, although my guess is that it did.

Human-rights groups and others claim broadly, and glibly, that torture and less-extreme coercion never, or almost never, extract useful information, because prisoners will make up false stories and confessions to stop the pain.

There is little or no empirical evidence bearing on this either way. But common sense suggests that blurting out the truth would be a better bet to stop the pain than concocting fabrications likely to crumble under pressure. And anecdotal evidence too extensive to detail here suggests that coercive methods have sometimes worked. The Israeli government has long said this with confidence. Remarkably, the highest Israeli court said the same thing even as it banned highly coercive methods in 1999, explaining that in a democracy "not all means are acceptable."

The worst thing about the Bush administration’s approach has not been the suffering it has caused KSM and a few other bad men. It has been the inexorable pressure on interrogators to brutalize lesser Qaeda functionaries, foot soldiers, and innocent joes mistakenly seized as terrorists, all the way down the slippery slope to Abu Ghraib and Bagram Air Force Base in Afghanistan, where two small-fry prisoners were reportedly tortured to death. This has shamed our nation and played into our enemies’ hands by staining our image abroad.

The best way to set things right would be for the next president to work with Congress to produce a comprehensive new legal regime that involves a number of changes.

To prevent interrogators from going too far, any order should detail in writing what methods can be used and for how long.

•The specific interrogation rules would no longer be established unilaterally by the executive branch but instead would be democratically legitimated through detailed congressional input reflecting voters’ value judgments.

• Congress should drop the proposal that it passed in February (and that Bush vetoed) to bar even mild interrogation methods that "threaten or coerce" a prisoner in any way–such as angry shouting and the like. It was appropriate to impose such rules on the military, as did the 2005 McCain amendment. But the CIA, which holds the most information-rich prisoners and has the most expert interrogators, needs more latitude, as detailed in my December 15, 2007, column.

• Congress should reaffirm that torture is always a crime and clarify the much-disputed definition. For U.S. law to bless torture under any circumstances would be an invitation for abuse and another disaster for America’s image.

• Highly coercive methods close to the line of torture–including waterboarding, which arguably crosses that line–should also be illegal, but with a narrow exception: Congress should authorize the president to order on a case-by-case basis specified coercive methods short of torture to pressure the rare captive who seems especially likely to have lifesaving information. To prevent interrogators from going too far, any such order should detail in writing what methods can be used and for how long. To provide political accountability, the president should sign such orders, share them with the intelligence committees, and publicly disclose their number.

Such a combination of an absolute ban on torture and a somewhat less absolute ban on other highly coercive methods has been suggested by a number of thinkers, including Philip Heymann, Juliette Kayyem, Benjamin Wittes, and National Journal‘s Jonathan Rauch. Philip Bobbitt suggests a variant to guard against official error and overzealousness in his magisterial new book, Terror and Consent: allowing coercive interrogation only when a special jury of ordinary citizens "can be persuaded that the detainee is in fact a terrorist with valuable information."

None of this is to deny that as a matter of morality, if "the stakes are high enough, torture is permissible," as Judge Richard Posner wrote in The New Republic in 2002. But so rarely (if ever) will it be knowable in advance that a prisoner has information that could avert catastrophe that anyone who resorts to torture should be deemed a lawbreaker whose protection must lie in public opinion, the president’s pardon power, prosecutorial discretion, the common sense of jurors, and the judgment of history.

John McCain, who was famously tortured in North Vietnam and has been a leading opponent of the Bush interrogation regime, seems to get that. Presented in 2005, by Newsweek, with a nuclear-bomb-hidden-in-New-York-City hypothetical, McCain cited Abraham Lincoln’s probably unconstitutional suspension of habeas corpus to save the union. "You do what you have to do," McCain said, "but you take responsibility for it."

This article appeared in the Saturday, May 3, 2008 edition of National Journal.