Evaluating Conventional Prompt Global Strike

Abstract

Arming Trident ballistic missiles with conventional warheads to strike “high-value” targets on a moment's notice would likely cause more problems than it solves.

At the request of Congress, in February 2007 the National Research Council of the National Academy of Sciences assembled a panel of experts under the sponsorship of the U.S. Navy to assess a controversial Defense Department proposal–rapidly converting 24 Trident II submarine-launched nuclear ballistic missiles into non-nuclear weapons. By adding conventionally armed missiles to submarines that previously carried only nuclear Tridents, the United States would be able to strike any point on Earth within about an hour of the decision to launch, without resorting to nuclear arms. This “conventional prompt global strike” capability would transform a leg of the nuclear triad from a never-to-be-used strategic deterrent into an active tool in the battle against terrorist organizations and WMD proliferators.

Prompt global strike is not a new concept. It has been advocated since the early 1990s; Defense described it as part of the “New Triad” in the 2001 Nuclear Posture Review; and the 2006 Quadrennial Defense Review specifically endorsed the “conventional Trident modification” idea. But legislators from both parties have questioned its merits, noting the inadequacy of intelligence to identify targets reliably on short notice. Opponents also have raised the issue of “nuclear ambiguity”–the unnerving possibility that Russia might misinterpret the flight of a conventional Trident as a nuclear attack and respond accordingly. Therefore, Congress charged the expert panel to consider the broader questions involved in pursuing Trident modification or other alternatives.1

When the panel issued its final report in August–dubbed “U.S. Conventional Prompt Global Strike: Issues for 2008 and Beyond”–it did not share Capitol Hill's skepticism and recommended the expeditious development, testing, and deployment of conventional Trident missiles. Unfortunately, the report barely touched on the question of intelligence requirements and dealt with the issue of nuclear ambiguity in a conflicted, inconsistent manner. So far, Congress appears unconvinced by the report's conclusions–and with good reason–but the debate continues. Future assessments should address more carefully what the August report neglects or treats equivocally.

Intelligence, uncertainty, and risk. The report envisions employing conventional Tridents against “fleeting targets” (i.e., elusive terrorist leaders or vehicles transporting weapon-related materials) whose detection would leave little time for deliberation before attack. Thus, the quality of intelligence required before acting becomes crucial to shaping the “rules of engagement,” the standing guidance about when (and when not) to shoot.

While more general, less time-sensitive intelligence reporting (called a “strategic warning”) would inform decisions about what targets to seek out, any decision to fire would require an “actionable warning”–specific, reliable information such as electronic or human intelligence identifying the location of a particular target at a particular time. According to the report, the decision to shoot would be made “perhaps in tens of minutes” after the receipt of an actionable warning; otherwise, the target could escape. The report maintains that these weapons would be used “only infrequently and on very special missions,” against only “the most important and time-urgent targets,” and recommends that because of their extraordinary nature, the authority to use conventional prompt global strike should be given only to the president. It does not outline, however, how the president would establish the reliability of each “actionable warning.”

Based on experience, there would be many such warnings. The pursuit of a high-value target produces a high volume of reports, each of uncertain merit. This flow of information imposes a burden of judgment upon the official charged with deciding whether to shoot–a particularly heavy burden if there is no opportunity to confirm the soundness of each initial report before action is required.

While it is difficult to predict how often conventional Tridents would be used, under sufficiently loose rules of engagement it is fair to reason that the real limits on how many shots are fired will be the number of “actionable” reports received and the number of weapons available. Already, strikes against fleeting, high-value targets have become commonplace in U.S. counterinsurgency operations, so much so that if each sighting of a “high-value” terrorist or insurgent leader in the tribal areas of Pakistan last year had generated a conventional Trident launch, the initial fleet-wide load of 24 weapons would have been expended by roughly October, or perhaps as early as April.2

While it is difficult to predict how often conventional Tridents would be used, under sufficiently loose rules of engagement it is fair to reason that the real limits on how many shots are fired will be the number of “actionable” reports received and the number of weapons available.

Reliable information about the results of airstrikes in Pakistan is elusive, but a detailed review of the outcomes of airstrikes in Afghanistan from late-2005 to July 2008 reveals a distinct pattern: While just 11 civilian deaths have resulted from strikes planned in advance, the remainder of the 556 civilian deaths in airstrikes were attributed to unplanned attacks on “targets of opportunity.”3 These deaths have put considerable pressure on the Afghan government, forcing the United States and its military partners to reconsider the rules of engagement.4

At times, the unintended consequences of even a single strike in error have proven intolerable. In July 1988, for example, the crew of the USS Vincennes, operating in the Persian Gulf, misidentified a civilian passenger jet taking off from Iran as an attacking warplane. The resulting shoot-down killed 290 people. Misidentification also can produce strategic setbacks. During the spring 1999 Kosovo conflict, U.S. B-2 bombers mistakenly dropped satellite-guided bombs on the Chinese Embassy in Belgrade, which the CIA had erroneously selected as a target. Three people were killed, and 27 were wounded; the CIA's role in future targeting was immediately curtailed. In the subsequent furor, student protesters laid siege to the U.S. Embassy in Beijing. Later, high-ranking U.S. officials offered their skeptical Chinese counterparts a formal apology and a detailed briefing on the nature of the error. These steps did not dissipate an atmosphere of deep mistrust between the two powers, and Chinese military modernization efforts accelerated.5

Reducing uncertainty and risk. While hunting terrorists, the U.S. government has typically employed a cautious approach based on an appreciation of the limits of its intelligence. For instance, while former Director of Central Intelligence George Tenet testified before the 9/11 Commission that the agency's pursuit of Al Qaeda leadership in the late-1990s involved “an aggressive human source collection effort–both unilateral and joint with liaison part-ners–and a vigorous development of technical collection within Afghanistan,” he later conceded that the resulting product was inadequate for rapid and reliable targeting. In his memoir At the Center of the Storm, he described it as “secondhand information relayed by our tribal assets or picked up in signals intelligence and analyzed days later.”

Likewise, former Defense Secretary William Cohen told the 9/11 Commission, “The information on [Osama] bin Laden and other Al Qaeda leaders was often from sources of questionable credibility, frequently fragmentary and packaged in inference, and ultimately of dubious reliability.” For these reasons, he added, “[T]he military gun was cocked for an extended period, but only once was the intelligence adequate to pull the trigger and launch strikes in an attempt to kill bin Laden or any other Al Qaeda leader.” Such caution was the result of awareness that a single miss might jeopardize the entire enterprise: “The decision to use force against a site at which bin Laden might be located required weighing the probability of successfully getting bin Laden. … Had we destroyed a compound and its inhabitants based on flawed or inadequate intelligence, international cooperation in tracking and seizing Al Qaeda operatives would have very likely diminished significantly.” For this reason, when potentially actionable intelligence was received, forward-based military assets were moved into position to strike while confirmation was sought.

Deliberative rules of engagement have proven successful, even against elusive terrorists. In May 2008, for example, a barrage of Tomahawk cruise missiles struck a cluster of buildings in Dusa Marreb, Somalia, killing an important Al Qaeda operative and his associates. An unnamed Pentagon official told the New York Times that the attack was “in the works for some time” and described to the paper the weeks-long process of tracking the target through multiple means of intelligence.6

Technological advances can also improve the quality of uncertain intelligence reports. Unmanned aerial vehicles (UAVs) can provide high-quality intelligence about a target's identity, and armed UAVs can conduct the attack against that target.7 In the last two years of fighting in Iraq, armed UAVs have enabled tighter integration between different forms of intelligence, airborne sensors, “trigger-pullers,” and weapons, reportedly producing many successes. Similar practices are now being employed along the Afghan/Pakistani border.8 Somewhat discordantly, “U.S. Conventional Prompt Global Strike” acknowledges the importance of forward-deployed special forces and UAVs in providing intelligence, but only in support of conventional Trident strikes. Because they have demonstrated their ability to perform the attack as well, UAVs and special forces render conventional Tridents moot.

That said, no form of intelligence can completely eliminate uncertainty, and no method of attack is free of risk. Even entirely successful and “clean” strikes against terrorist targets can be provocative, as illustrated by the sharp reactions to raids by U.S. Special Forces within Pakistan and Syria.9 Risk can be reduced to acceptable levels, however. To do so, a commander in pursuit of elusive targets must seek to minimize both the probability and consequences of erroneous strikes. The probability of error declines with attacks supported by better intelligence, while the consequences of error decline with subtler, less dramatic forms of attack.

To its credit, the national research council report clearly favors measures that ease concerns about nuclear ambiguity–the possibility that a nuclear-armed adversary could mistake a conventional ballistic missile launch for a nuclear strike.

The alleged role of execution time. According to “U.S. Conventional Prompt Global Strike,” an essential ingredient for success in striking fleeting targets is execution time–i.e., how quickly an attack can be performed from a standing start. To illustrate the point, the report mentions two failed strikes: the August 1998 cruise-missile attack on a camp near Khost, Afghanistan, where bin Laden and several associates were planning to meet, and the March 2003 airstrike against Saddam Hussein's compound at Dora Farms, popularly referred to as the opening shot of the Iraq War.

Indeed, some accounts written by former government officials speculate or assert that excessive execution time foiled the bin Laden strike. In Against All Enemies: Inside America's War on Terror, former White House counterterrorism adviser Richard Clarke hints that bin Laden might have been tipped off during the two-hour flight of the cruise missiles, thanks to the coastal surveillance activities of the Pakistani Navy.

But journalist John Diamond's thorough and judicious assessment of the attack in his recent book, The CIA and the Culture of Failure: U.S. Intelligence from the End of the Cold War to the Invasion of Iraq, reveals a different, largely overlooked set of causes that began with the arrest of Al Qaeda operative Mohammed Sadeeq Odeh. Closely involved in the bombings of the U.S. embassies in Kenya and Tanzania earlier that August, Odeh planned to attend the same meeting in Afghanistan as bin Laden. Instead, he was detained en route at the Karachi airport in Pakistan, creating a series of informational ripples that would have been hard for bin Laden and his associates to miss: Odeh's failure to reach Afghanistan; the word of his arrest, presumably communicated by his traveling companions who were not similarly detained; and Pakistani newspaper and television accounts of his arrest and interrogation. Some of these same news reports claimed that the United States wanted to enlist Pakistan's cooperation in ambushing bin Laden in Afghanistan.

Under these circumstances, it is difficult to imagine that the Khost meeting could have gone ahead as planned. Yet as Diamond notes, the attack planners at the White House did not take these developments into account. Nor did they include the FBI, which interrogated Odeh five days before the bin Laden strike, in their highly secretive deliberations. Thus, the attack's failure seems to owe more to a restricted flow of information than to the two hours needed for the cruise missiles to reach their target.

There is even less reason to believe that execution time influenced the outcome of the 2003 Dora Farms attack against Saddam Hussein. A subsequent reconstruction of the Iraqi leader's movements by U.S. Joint Forces Command reportedly found that he had been nowhere near the location at the time.10 In hindsight, the role of overconfidence in questionable intelligence is all too apparent. In an initial report, a CIA officer deemed a human intelligence finding that purported to relay Saddam's movements as “99 percent certain,” because “[n]othing is 100 percent.” Satellite photos appeared to corroborate the report, showing an underground bunker at the site. But by the time the same intelligence officer accompanied Tenet to brief President George W. Bush, his confidence had fallen to 75 percent. Nonetheless, Tenet assured Bush, “This is as good as it gets.”11

Gen. Michael Moseley, the war's air commander, was told that the CIA was “99.9 percent sure” about the presence of the intended target at Dora Farms. After Moseley informed Washington that he had the necessary forces to execute the mission, he received the president's order to strike. The attack hit its aimpoint–a bunkerless “empty field.”12 But immediately after the strike, the flow of faulty intelligence continued, with U.S. forces receiving reports from a human source that Saddam had been removed from the wreckage on a stretcher.13

Non-nuclear strategic warfare. To its credit, the National Research Council report clearly favors measures that ease concerns about nuclear ambiguity–the possibility that a nuclear-armed adversary could mistake a conventional ballistic missile launch for a nuclear strike. In particular, the report rejects near-term, land-based alternatives to the conventional Trident because such weapons would need to pass over Russian airspace to reach most points in the Eastern Hemisphere. For the same reason, it proposes that the ocean patrol areas for conventional Trident-armed boats be located further south than would be ideal for the nuclear deterrence mission.

But its position is complicated–if not completely undermined–by a pattern of equivocation about the desirability of additional conventional prompt global strike systems beyond non-nuclear Tridents, some with the ability to strike hardened targets deep in foreign territory. At one point, the report refuses to rule out the use of conventional prompt global strike weapons against Russia and China. The risks, it declares, are “sufficiently low and manageable that, while their existence needs to be taken into account–and the use of [conventional prompt global strike] even foregone in certain cases because of concerns about escalation–they do not constitute a reason to forgo acquiring the capability.”

This sanguine attitude about “escalation” to nuclear war certainly runs afoul of concerns about nuclear ambiguity. If a ballistic missile overflight of Russia en route to third-party targets would be unacceptable, it is difficult to understand how actual attacks on Russian or Chinese targets with the same weapons could be acceptable. The paradox leads to an extraordinary line of argument in the report: “Of course, any attack on a nuclear-armed country is a very serious matter, simply because of the possibility of escalation, but the probability of a nuclear response to such a conventional attack is surely lower than the probability of a nuclear response to a nuclear attack.” In other words, conventional prompt global strike attacks on nuclear-armed powers are acceptable because they are less dangerous than preemptive nuclear attack–an argument that could justify almost any act imaginable.

What interests are worth defending in such a fashion? If the protection of a particular asset, say, a satellite constellation, requires gambling with millions of lives, is that asset worth protecting? It seems more reasonable to pursue methods of defense that do not rely on preemptive strikes deep inside the territory of a nuclear-armed nation.

It is already apparent that “U.S. Conventional Prompt Global Strike” will not be the last word on the subject. Congress has instructed Defense, in consultation with the State Department, to conduct a further review of conventional prompt global strike. The review, scheduled for delivery in September 2009, is to include an investigation of the nuclear ambiguity problem, types of targets, and intelligence-gathering requirements. But it is far from clear that these issues can be properly satisfied. And if an appropriate mission and concept of operations cannot be devised for conventional prompt global strike weapons, Congress and the Obama administration should not rush to deploy them.