The American ships and planes that carried military supplies to Israel during the 1973 Arab-Israeli war did not all return empty. Aboard one was a great prize of espionage—a captured Russian super-weapon called the ZSU-23-4, which was known to intelligence agents as Shilka. What exactly a Shilka was, and what it might do, had long troubled the Pentagon. Although it looked like nothing more than a small tank, Shilka was in fact a radar-directed, computer-controlled anti-aircraft cannon. If, as rumored, it could destroy even low-flying aircraft with ease, Shilka might make Soviet armored columns nearly unstoppable.

The weapon, taken by Israeli soldiers from an Egyptian tank division, was moved to a mountain gunnery range near Fort Bliss, Texas. There, in a series of tests code-named Hitval, Shilka’s performance was evaluated. Fortunately for U.S. pilots, the results were not impressive. Shilka’s fire-control computers were inaccurate and slow. It rapidly exhausted its ammunition, and was very difficult to reload. But most important, it proved unable to hit maneuvering targets. It could threaten only aircraft flying straight, predictable lines—and no aircraft is likely to do that in combat.

Shilka, then, was not the decisive super-weapon many had feared. But, looking it over, U.S. Army officials were envious, nonetheless. The Army badly needed a new anti-aircraft gun to protect its tanks and troops at the front lines. And the Army badly wanted what Shilka had in abundance: the glamour of high technology. Since World War II, nearly all the expensive, exotic weaponry had gone to the Air Force and the Navy; the Army had been consistently frustrated in its desire to obtain what around the Pentagon is called “ultra” or “cosmic” equipment. The Army had been trying to change that for years, and during the 1970s it won congressional funds for a jet-engine-powered tank (the M-1); a laser-guided artillery shell (the Copperhead); an anti-aircraft missile patterned after the anti-ballistic missile (the Patriot); a $1.6 million armored, “air-droppable” bulldozer (the ACE); a helicopter that costs at least as much as a supersonic fighter plane (the AH-64 Apache); a helicopter-borne radar system modeled after the one on the AWACs surveillance plane; and many other sophisticated weapons. Pondering Shilka’s computers and radars, Army officials began to think that something along the same lines would suit their high-technology campaign perfectly.

That something is now going into production: a Shilka-inspired anti-aircraft vehicle called the Sergeant York but better known as Divad (for “division air defense”). Divad looks and drives like a tank, and its armament—medium-sized cannons—is not unusual. Everything else on it is. Atop Divad’s turret are two radars adapted from the F-16 fighter. Inside is a computer network designed to track aircraft, “assess” their “threat levels,” predict their movements, and aim the guns—all automatically. (Divad requires “essentially no skill to operate,” according to Keith Harder, technical manager for Ford Aerospace, the Ford Motor Company division that is building Divad. “About all they have to do is turn it on and pull the trigger,” he says. A light on the control panel flashes “FIRE NOW.”) Deeper inside Divad is a wide array of “bites”—“built-in test equipment”—designed to monitor the radar and computers and provide instant  analysis of anything that goes wrong. (Bites are, themselves, linked to computers.) Other components include computers that resist radar jamming; “IFF interrogators,” said to be able to determine whether an aircraft is friend or foe; a laser range-finder and an infrared night-sight to back up the radars; and air-conditioners to keep all this technology from overheating. Army documents refer to Divad not as a gun or a vehicle but as a “launch system.”

These devices will make Divad the most sophisticated piece of equipment ever to roll onto a battlefield, and it is priced accordingly. At a minimum of $6.8 million each. Divad will cost nearly three times as much as the M-1 tanks it is designed to defend. The full production budget will be at least $5 billion. Yet Divad has progressed from drawing board to assembly line nearly as quietly as its Russian counterpart. At this writing, to the best of my knowledge, the Divad project has never been discussed in major newspapers except for items about Ford’s contract awards. Likewise, it has been debated in only two congressional hearings and in those only briefly.

While the public may be ignorant of the project, Divad is notorious inside the Pentagon. “We’ve finally found a way to make a gun more expensive than a missile,” one recently retired colonel has said scornfully. More troubling than the expense, however, is the fact that Divad’s electronics, though of much higher quality than Shilka’s, are not much more effective. Like Shilka, Divad cannot hit an aircraft that is flying evasive maneuvers.

How Divad came to be is a case study in the modern high-tech military, where drastic increases in defense spending are often accompanied not by increases in U.S. fighting strength but by declines. In the rush to make Divad an “ultra” weapon, Army planners brushed aside a much cheaper and simpler system that, experience had proven, would do a better job of destroying enemy aircraft. Once the “ultra” approach as chosen, internal Pentagon politics and the influence of defense contractors became the driving forces of the project. A super-expensive Divad built by Ford Aerospace became, to the Army, an end in itself, and whether or not the system worked was reduced to little more than an annoying side issue.

Hardly anyone disputes the army’s need for a new anti-aircraft weapon. That need dates back to the 1950s, when, according to Lt. Gen. James Maloney, head of the Army’s Air Defense Center, the service “made a conscious decision to eliminate division air defense.” (“Division” in this sense refers to tanks and troops actively engaged in battle.) The Army was enamored of surface-to-air (SAM) missiles, which it developed at considerable expense through the 1950s and 1960s. But SAMs turned out to be so complicated and hard to set up that they could be used only to defend rear-area targets such as supply depots; they were no help to soldiers at the front. Moreover, all SAMs—U.S. and Soviet—proved to have a technological blind spot: they did not work against aircraft flying low to attack troops.

Meanwhile, the Army had grown accustomed to operating with complete air superiority. In Korea, it faced only occasional air attacks, and in Vietnam it faced none (the Viet Cong had no airplanes). Any future combat in Europe would be different, however. As the 1960s progressed, the Soviet Union began building strike fighters and fielding teams of “forward air controllers,” whose purpose was to locate small targets such as tanks and to direct attacking aircraft to them. In response, the U.S. built a hand-held SAM of dubious utility called Redeye, which was slower than most of the aircraft it was supposed to catch, and small numbers of a somewhat mobile SAM called Chaparral.

Luckily, the Army began to have reservations about the all-missile approach. Great Britain’s military had none. The British fleet that sailed to the Falkland Islands was defended almost exclusively by SAM missiles; British planners had assured their admiralty that sophisticated electronics make guns and gunners obsolete. The result of the fighting showed the extent to which high technology falls short of its promise when it is tested in combat. All four British warships sunk off the Falklands were brand-new air-defense ships heavy with expensive SAMs and radars, and all four were sunk from the air. One, the destroyer Sheffield, was hit by a guided missile, but the others were victims of unguided “dumb” bombs dropped by airplanes passing directly over the decks. As the vulnerability of British ships to low-flying aircraft became clear, London sent out an emergency call to the international arms market for cannons, which were bolted with furious speed onto the decks of ships scheduled to sail if the crisis continued. BBC-TV beamed back footage of British marines setting up their machine guns on every free spot of deck space on the carrier Hermes, trying to compensate for the fleet’s unrealistic high-tech battle plan.

When British soldiers went ashore on the Falklands, they suffered similar handicaps; they were equipped with two types of SAMs but no anti-aircraft guns. According to knowledgeable sources, Blowpipe, a hand-held SAM, failed to make a single hit, despite dozens of firings. Rapier, which is mounted on a platform, was more effective, but it required so much setting up and calibrating that it often wasn’t functional when attackers appeared. When Argentine planes struck British landing craft in a bay near Fitzroy, sinking the assault ship Sir Galahad, Rapiers on shore were supposed to be providing cover; they had been landed first for this purpose. But the Rapiers were not yet set up, and could not be fired. In all, the British claim to have shot down about seventy-five Argentine airplanes, two thirds of them by Harrier fighters in air-to-air battles. Thus the hundreds of advanced SAMs that the British took to the Falklands destroyed only twenty-five planes—planes that were twenty years old and had no radar jammers or other SAM-foiling devices.

When U.S. army officers, unlike their British counterparts, awoke to the drawbacks of SAMs in the 1960s, their first response was to build a light anti-aircraft cannon called Vulcan. Unfortunately, Vulcan outlived its usefulness with the advent of armored aircraft, which its 20mm shell is not powerful enough to destroy. At about the same time the Shilka was captured, the U.S. began building an armored attack aircraft, the A-10, and the SOivets began developing one, the Sukhoi 25; by the mid-1970s, Divad had become a high-priority project.

Although the Army had faced up to the electronic limitations of SAMs, it seemed to deny that there was any limit to what engineers could do with a gun. “Divad started out as a reasonably simple system, but pretty soon we were adding every bell and whistle you could think of,” says one weapons designer involved in its planning. “Bells and whistles” is Pentagon slang for extravagant frills.

Divad’s complexity was provoked by the Army’s dream of mounting a technological response to the threat posed by “all-weather” strike fighters. “All-weather” aircraft have radars and electronic sensors that supposedly enable them to attack tanks in the dead of night or through clouds; giving Divad radars and computers supposedly would enable it to fire back under those conditions. “A person may be able to aima gun during the day, but he can’t aim at night or in fog,” Lieutenant General Maloney has said. “We need automatic fire-control for those times.”

Military history, however, suggests that at night or in fog there will be nothing for Divad to shoot at. Even the U.S. Air Force, with the world’s most advanced electronics, seldom flies missions in bad weather, because it is difficult and dangerous to do so even in the absence of enemy fire. Two of Britain’s advanced Harrier warplanes crashed in fog off the Falklands without ever being shot at. In Vietnam, the Air Force’s cost-no-object, all-weather attack airplane, the F-111, which is crammed with radars and computers, rarely operated in bad weather, and was ineffective when it did. Combat photographs showed that the accuracy, called “circular error probable,” or CEP, of the F-111’s automated bombing system was three quarters of a mile. That is sufficient to threaten large, stationary targets (rail yards, for example) but not small, constantly moving tanks. When tanks are the target, the “destruct radius”—the CEP required—of a 500-pound bomb is fifteen feet. “Once in Vietnam I bracketed a T-54 tank on both sides with 500-pounders,” the pilot of an F-4 fighter modified for ground attack told me recently. “The guys inside probably didn’t hear anything for a week, but they were still able to throw the tank into first and run for the tree line. I didn’t get them.”

Since Vietnam, all-weather devices have improved, but not to any degree that would threaten tanks. This past July, four F-16s, equipped with state-of-the-art avionics, staged a demonstration of close-quarters bombing at Ford Bragg, North Carolina. The aircraft were operating under the ideal conditions of clear weather and no opposing fire; nevertheless, three out of the four planes missed their targets, two by half a mile. At about the same time, a writer in Aviation Week, the defense industry’s bible, gave an account of his “test ride” on the most elaborate all-weather aircraft ever constructed, the $40 million F-15E “Strike Eagle” version of the F-15 interceptor. On its best pass, under ideal conditions, the F-15E, bearing nearly every electronic-guidance and sensing mechanism that exists, missed its target by more than a hundred feet—still, by anti-tank standards, a gutter ball. The main Soviet all-weather aircraft, the Fencer, resembles an F-111 and has similar terrain radars. Most defense analysts find it hard to imagine that the Fencer could be any more effective than the F-111, the F-15, or the F-16, given the inferior state of Soviet electronics technology. If they are right, then the primary all-weather threat to U.S. tanks will be gutter balls missing by hundreds, if not thousands, of feet.

The one new electronic system that does pose a threat to Army tanks, many analysts agree, is a night-sight called Flir, for “forward-looking infrared.” Flirs allow pilots to see a small slice of ground in front of them almost as if it were day. Flirs are unreliable in high humidity, however; they work best on clear, dry nights. (This is why the Army, which once touted the Flir-equipped AH-64 helicopter as “all-weather,” has since changed the craft’s billing to “adverse-weather.”) Aircraft bearing Flirs can be countered by a simple expedient that avoids costly radars: anti-aircraft guns with Flirs. In a Flir-to-Flir duel, ground gunners would have a distinct advantage. Aircraft flying against the cold, empty background of the sky provide an excellent thermal contrast, sharp and easy to see. Vehicles prowling amid the clutter and retained warmth of the ground, on the other hand, provide only a small thermal contrast; heat “camouflages” them against infrared detection the same way trees and colors camouflage them against visual detection.

Eventually, Divad’s “All-Weather” rationalization lost respect around the Pentagon, but a new, presumed technological threat arose to take its place—missile-firing helicopters. Today, these helicopters are Divad’s official reason for being: “A majority of the targets Divad will engage will be helicopters,” Col. Charles Clarke, a Divad project officer, recently told me. “The electronics are being optimized to meet the helicopter threat.”

The Army’s claim that it needs a $6.8 million radar cannon to shoot down helicopters strains credulity, because experience has shown that even a $600 rifle can do the job. The Army’s report of the number of helicopters it lost in Vietnam is 4,643; nearly all of them were lost not to SAMs or automated cannons but to rifles and machine guns—weapons that the Viet Cong carried on their backs.

Simple small-arms fire destroyed so many helicopters in Vietnam because helicopters are by nature extremely delicate. Many of their critical parts are exposed, and cannot be armored heavily, because weight in helicopters is at a premium. Helicopters also move slowly, compared with airplanes, and cannot make fast, evasive turns. Given their vulnerability to small-arms fire, they are even more vulnerable to the larger-caliber weapons of a mechanized division.

Only on rare occasions during the Vietnam War did U.S. helicopters fly against anything resembling the large-caliber anti-aircraft defenses present in NATO and Warsaw Pact armies. When they did, the results were catastrophic. Operation Lam Son 719, an attempt in 1971 to shut down Laotian supply lines, saw U.S. helicopters pitted against large-caliber machine guns and light cannons similar to Vulcan. In two months, 107 helicopters were destroyed and 608 damaged, many seriously; the rate of helicopters lost per sortie in Laos jumped to twenty times the rate for helicopters flying over the Viet Cong in South Vietnam, Pentagon figures show.

Great vulnerability is by no means limited to U.S. rotary craft. The Soviet Union has been using its most advanced attack helicopter, the Hind-D, in Afghanistan; a number of them have been shot down by Afghan rebels armed only with machine guns and with tiny, Redeye-like rockets.

The Army maintains, however, that U.S. helicopters have recently become more deadly and more “survivable,” because of the addition of electronic weapons, known as “standoff missile”; therefore, it assumes, Soviet helicopters likewise will be improved. Cobra attack helicopters and the planned AH-64 Apache have anti-tank standoff missiles that, in theory, can be fired from tree-top level three or four kilometers away, outside the range of small-arms fire, and guided to their targets for direct hits. Like all-weather ground scanners, if such weapons worked under realistic conditions, they would pose a grave threat. Many defense analysts doubt that they will work, however. Helicopter gunners must first spot a target with their eyes; even the most electronically advanced helicopters require this. Most tests of helicopter standoff missiles have been conducted over flat desert terrain, where three- to four-kilometer sight line are possible. The rolling hills and tall trees of Europe, however, present such steep “look angles” that helicopters would have either to draw close to their targets or to rise high in the sky in order to get a sight line; either way, their standoff advantage would be lost.

When standoff missiles are tested, helicopter gunners usually know the location of their targets in advance. This avoids the greatest problem of air-to-ground combat: figuring out where to shoot. A tank, while looming large to someone standing beside it, is little more than a dot to a helicopter gunnery officer four kilometers away, and often a camouflaged dot at that. Colonel Clarke, who once commanded a Vulcan battery, recalls having ordered his unit to camouflage itself as part of a drill; he then flew over it in a helicopter. “I couldn’t see them at all, even though we were flying right by and not under any kind of duress from ground fire,” he says. “I just couldn’t see them.”

Yet, as Divad’s planning progressed, the Army added more and more bells and whistles designed to counter hypothetical helicopters performing with technological perfection. At work was a self-fulfilling prophecy: having convinced itself that its own missile helicopters were the ultimate threat to Russian tanks, the Army was obliged—especially when requesting funds from Congress—to quake at the prospect of missile-bearing helicopters on the other side. The Army declared that future helicopters would be able to launch missiles in fifteen seconds. Cannon shells take seven seconds to fly to the maximum standoff range of four kilometers; that would leave eight seconds for Divad to spot a helicopter, slue its turret around, compute, aim, and fire. So the central technological goal of Divad became, according to project participants, “eight seconds.” Obviously an “ultra” system would be required. “Eight seconds was all we ever heard about; it meant we had to push everything right to the outer limits,” a designer of one competitor for the Divad contract says.

“Optimizing” Divad for all-weather and eight-second performance meant that the bulk of the program’s efforts—and expense—would be focused on threats that were marginal if not imaginary. Some of the problems posed were almost comical. It was found, for instance, that radar could track a fast-moving helicopter high in the sky much more easily than it could a hovering helicopter low to the ground. (Radar waves traveling near the ground bounce off trees, rocks, and other “clutter” to create distortion similar to the distortion that spoils FM radio transmission.) Maj. William Gardepe, a Divad project officer, explained how electronics engineers had labored to overcome this obstacle. I suggested that aiming at a stationary helicopter is so easy that almost anyone can do it—without technological breakthroughs. He gave me a hurt look and said, “But that was the technical challenge.”

In 1977, after looking over proposals for electronic guns from Ford, General Dynamics, General Electric, Raytheon, and Sperry-Rand, the Army awarded development funds to Ford and to General Dynamics. Each was to build a complete Divad prototype, and a “shootoff” would determine the winner of the production contract. In the spring of 1980, the shootoff, officially called DT/OT II, began. Quickly it became clear that abstract “technical challenges” had driven out real, proven threats as concerns of Divad planners; neither weapon, it seemed, would be particularly effective against regular airplanes dropping dumb bombs on sunny days.

Through the course of DT/OT II, the Divad contenders shot down two F-86 fighters, five Huey helicopters, and twenty-one small-scale drones. Both outperformed Shilka. But the Army never tested them in the one critical area where Shilka had failed: neither gun shot against a maneuvering target. The F-86s were hit while flying straight, predictable courses several hundred feet above the ground. The helicopters were hit while flying lazy, level courses or hovering. According to Major Gardepe, the sharpest evasive maneuver any target flew against Divad during the shootoff was a turn at two Gs—that is, a turn generating twice the force of gravity. By aircraft standards, this is quite modest. Even an average-sized sedan can generate seven tenths of a G in turns on freeway ramps.

Divad was shooting, in the words of a retired Air Force colonel and fighter pilot, against “extremely cooperative targets.” He explains: “During wartime, attack pilots will be in constant turning flight, jinking [making unpredictable wriggles] and changing altitude.” A pilot of the A-10 attack plane says he has flown seven-G turns just 100 feet above the ground, and could use three-to-five-G maneuvers during an attack run. “There’s no way I’d fly straight and level over tanks, and neither will anybody else,” he says.

For its part, the Army says it has determined that attack planes cannot use high-G maneuvers during attack runs, and therefore has dismissed the problem from consideration. (This same wishful thinking handicaps other anti-aircraft weapons. At Fort Bliss several months ago, I watched Vulcan gunners train. They fired at a drone flying sweeping, graceful figure eights hundreds of feet high, a target unlike any they would ever encounter in wartime. Nearby, gunners were practicing with the Chaparral SAM and an improved hand-held SAM called Stinger. They fired at drones tracing a gradual arcing course, every drone flying the same course through the same patch of sky.)

At any rate, there was no point in testing Divad against maneuvering targets, the Army says, because the Army already knows that Divad can’t hit them. “Of course it will miss if the target jinks,” Lieutenant Genral Maloney says. “No computer can handle a jinking target.” This is a function not of the quality of technology but rather of technology’s limits. Radar and computers can with great accuracy determine where an aircraft is; they cannot determine where it’s going. For any gun to hit a moving target, it must fire in advance of the target’s path (“lead”), the same way a quarterback must throw to where a receiver is going, rather than to where he is. Divad’s electronics track the “history” of an aircraft, assume that the craft will continue on that course and set the guns in advance. If the aircraft keeps flying straight, its destruction is likely. If it makes an unpredictable jink, Divad will miss.

Planning Divad, the Army bypassed a vastly cheaper system that can outwit maneuvering aircraft: the human eye. “The eye is much better at anti-aircraft gunnery than any automatic system,” says a Pentagon analyst who has helped design several current U.S. weapons system. All of the low-flying aircraft destroyed by ground guns in World War II were victims of visual aiming, and most attack planes of the time could make turns in excess of two Gs. The five British Harriers shot down over the Falklands (three others were lost to accidents) were hit by extremely simple, visually aimed cannons. Most revealing, though, are statistics from the Vietnam War. According to the Defense Department, 91 percent of the modern, high-performance U.S. jets lost over north Vietnma were shot down by guns, most of which were visually aimed; 5 percent were hit by SAM missiles, and 4 percent by other fighter aircraft. Practically all the jets lost over South Vietnam, Laos, and Cambodia were hit by visually aimed guns. All of the helicopters lost in Southeast Asia were hit by either guns or bazooka-like rockets with visual sights. Viet Cong and North Vietnamese gunners were assisted, when assisted at all, by a simple “range only” radar that told approximate distance to targets but did not track aircraft or aim guns.

It may seem preposterous that the eye can be more effective than advanced electronics. But the eye, in this case, is better informed. Radar “sees” an aircraft only as a blip, a circular point without features. When a radar-computer-system tries to guess where an aircraft is going to “lead” its gun, the system can rule out only destinations that are ballistically impossible (for example, an airplane traveling north cannot instantaneously travel south). Given that an aircraft such as the A-10 can turn 20 degrees though three dimensions and move 500 feet in a single second, from any given point an aircraft has thousands of possible destinations, reducing the computer’s chance of a correct guess to near zero.

Eyes, on the other hand, see not a featureless blip but a complete flying machine. Low-level aircraft are close enough to allow gunners to observe the angle of their wings, the attitude of their flaps, and so on. Form these clues, an experienced gunner can narrow down an aircraft’s path to a few possibilities, instead of thousands. If a gunner sees a banking airplane’s far wing roll up, for instance, he knows it is coming toward him—something a computer would not know.

Army officers outside the procurement hierarchy concede these difficulties in private, and say that in actual combat, Divad’s radars and computers will probably be turned off. Even Lieutenant General Maloney agrees that “under some conditions” Divad gunners would find visual aiming more effective than electronic aiming. He reiterated that visual aiming would work during the day but not at night or in bad weather. It remains to be seen how, if Divad’s electronics aren’t accurate during the day, they will become accurate at night.

Following the DT/OT II shootoff, Ford had a problem. Although neither Divad contender had performed particularly well considering the extreme expense involved, Ford’s gun did far worse than its competitor’s, according to informed sources. The precise results of the test are classified, but I have seen them and can say that Ford’s Divad prototype destroyed fewer than half as many targets as the General Dynamics gun; the longest range of a direct hit by the Ford prototype was a little more than half the longest range of a direct hit by the General Dynamics prototype; the Ford gun never achieved a direct hit on an airplane drone; it was able to hit only helicopters, the easier target.

Apparently, Ford had anticipated this complication and had prepared for it. On Ford’s team were four recently retired three-star Army generals who had many friends in the Pentagon: Lt. Gens. Eugene D’Ambrosio, Robert Baer, Howard Cooksey, and C. J. LeVan. Both D’Ambrosio, who retired from the Army in May of 1980, and Baer, who retired in June of 1980, had been deputy commanders of DARCOM, the Army’s weapons-procurement branch. D’Ambrosio was now chairman of the board of Day & Zimmermann, a Ford-Divad subcontractor, and Baer was vice president of XMCO, a consulting firm under contract to Ford. Cooksey, a consultant under contract to Ford, had been, before his retirement, in December of 1977, deputy chief of staff for research, development, and acquisition, the Army’s most influential high-technology post. LeVan works for a defense consulting firm called R & D Associates, which is, in turn, under contract to Ford Aerospace. His last assignment before his retirement, in June of 1978, was as director of plans and policy for the Joint Chiefs of Staff. Before that, he had been commander of the Air Defense Center. It was under LeVan’s direction that the “eight-second” requirement and others for Divad were written. According to Pentagon sources, LeVan is one of the best-connected retired officers in Washington, and was tireless in working his contacts on Ford’s behalf. LeVan, however, denies performing any “marketing” tasks, saying he only gave technical advice to Ford.

While it may seem that the results of a full-scale shootoff using live ammunition and flying targets would be conclusive, beyond the power of even generals to alter, in the modern Pentagon nothing is final until submitted to a computer. So when the results of DT/OT II were transferred to the Army’s Ballistics Research Laboratory for computer interpretation, Ford’s hopes were revived.

Soon, mysterious things began to happen. The BRL ruled that explosions of proximity-fused rounds—“flak shells” that go off in the general vicinity of the target—would be treated as direct hits, even though some had missed by as much as forty-five feet. Next, proximity explosions were classified as ensuring “kills” of their targets, though it is believed that armored aircraft such as the U.S. A-10 and the Soviet Sukhoi 25 would not be destroyed by far-off flak. Finally, all proximity-shell firings by the General Dynamics gun were disqualified, on the grounds that General Dynamics had not used a regulation fuse.

Exaggerating the effectiveness of proximity-fused rounds and then disqualifying all such rounds fired by the General dynamics contender left Ford as sole beneficiary of the BRL’s “interpretations.” Thus, by the time the BRL’s computer “lethality model” was sent back to the Pentagon, the results of the shootoff looked approximately even.

The BRL report put Ford back in the running, but the company remained in a weak position. Its gun fired 40mm shells, while the General Dynamics gun used 35mm ammunition. This seemingly trivial difference had caused Ford great heartache from the beginning of the Divad program. Divad, intended primarily to fight Soviet armies along the NATO Central Front separating East and West Germany, was supposed to be “NATO-interoperable.” That meant using the ammunition that NATO guns use, and the dominant anti-aircraft NATO caliber is 35mm. To choose Ford, the Army would have to violate its own principle of cooperation.

That this dilemma existed at all is evidence of pressure from contractors and of the Army’s weak response. In 1977, the Army told Congress that it wanted to guarantee NATO interoperability by specifying 35mm guns on all Divad contenders. Immediately the lobbying began. Ford had a marketing agreement with the Swedish firm Bofors, a maker of 40mm but not 35mm cannons; while Ford could have switched to a 35mm weapon for Divad, the potential profits from a 40mm weapon were higher. Likewise, General Electric, which manufactures 30mm cannons, stood to make more money if it was not required to switch. Soon the Army was in retreat. Department of Defense lawyers, the Army pleaded to Congress, had advised that specifying the caliber of Divad’s gun would be “anti-competitive” and could lead to lawsuits—“the most ludicrous excuse I’ve ever heard,” a high-ranking Pentagon official has told me. When the final Divad requirements were issued, they called for a gun “in the 30mm to 40mm range.”

Even as the results of the shoot-off were being transformed by computer analysis, the requirements for inter-operability with NATO were also changing. According to a knowledgeable informant, when the Army’s Source Selection Evaluation Board convened to choose between Ford and General Dynamics, it was told by Divad project officers that Ford’s gun would be much more interoperable, because the NATO countries had approximately 1,500 40mm weapons like Ford’s, but only 300-and-some 35mm weapons like those of General Dynamics.

These figures must have struck any informed Army officer as peculiar. West Germany, for instance, had just completed a conversion to 35mm weapons—a conversion that was the subject of considerable publicity that debate within the military community—and was widely known to have more than 300 35mm weapons in its inventory. The board was told that West Germany had 624 40mm weapons but only 200 of the 35mm variety. The true figures, according to West Germany’s Armament Sector for Defense Materiel, were 180 40mm weapons and 432 35mm weapons.

West Germany, it turns out, is not the only country converting to 35mm anti-aircraft weapons; the trend is widespread throughout NATO. Yet the selection board failed to take trends into account, and considered only the 1980 figures (and even those weren’t correct). By 1985, there will be almost twice as many 35mm weapons in NATO as 40mm weapons, according to informed Pentagon and arms-community sources—figures very different from what the selection board assumed. Already in 1982, no NATO country uses 40mm anti-aircraft weapons along the Central Front. One month after the 40mm Divad was chosen, necessitating a new set of NATO supply lines, Deputy Secretary of Defense Frank Carlucci issued a memorandum to all service chiefs calling for “armaments cooperation with our NATO allies,” and especially for “interoperability and standardization so we can better fight as an alliance.”

Many army officials must have known, of course, that the BRL report was faulty and that the NATO statistics were fudged; the true NATO statistics should not have been too difficult for military officers to obtain. So why did the Army go to such lengths to turn the case in Ford’s favor?

A powerful force seems to have intervened. As the date for the selection board’s meeting neared, in the early months of 1981, Ford motors issued its 1980 annual report, declaring a $1.5 billion loss—the largest corporate loss in U.S. history. Some Pentagon officials believe that, around this time, Ford approached the White House and asked for financial aid. President Ronald Reagan, however, had campaigned against the Chrysler bailout and bailouts in general, so Ford presumably would have been told that direct aid was out of the question. The White House seems to have agreed, instead, to help Ford indirectly: publicly, it would work to advance Ford’s request that Japanese auto manufacturers accept voluntary import quotas; privately, it would ensure that Ford would receive a share of Reagan’s huge increases in military spending—a kind of defense build-up bailout. The Pentagon officials surmise that the White House, without specifically naming any project, communicated to Defense Secretary Caspar Weinberger its desire that he look favorably on Ford’s defense-contract applications. In the winter of 1981, Ford, whose defense business is relatively small, was in the running for just one major contract: Divad. Army procurement officials were anxious to please the new White House (and not offend a new President, who offered them seemingly unlimited budget increases), but they were also worried, with some justification, about losing a supplier if Ford went bankrupt. In May of 1981, Ford’s Divad was chosen over the General Dynamics rival.

According to industry sources, General Dynamics was stunned. General Dynamics has a well-organized lobbying network, and under normal circumstances it would surely have created a scene in Congress. But, as it happened, early 1981 was an awkward time. The General Dynamics Electric Boat subsidiary was deep in an acrimonious debate with Adm. Hyman Rickover about who would take the blame for severe cost overruns in the Trident submarine program. Stories critical of the General Dynamics shipyards were appearing in newspapers all over the country; John Lehman, the secretary of the Navy, announced that he would have future submarines built overseas, if necessary, to escape from Electric Boat’s clutches. Meanwhile, General Dynamics was negotiating to buy the Chrysler subsidiary that builds the M-1 tank, a lucrative project offering one of the steadiest profit flows in the defense industry. (Chrysler was forced to sell; it needed cash.) Strictly speaking, the M-1 plant was not a Chrysler possession; it was a “GOCO” facility (government-owned, contractor-operated), and could not be transferred without the government’s permission. If General Dynamics had embarrassed the Army over its Divad decision, permission might not have been granted.

Thus, Pentagon sources say, General Dynamics seems to have been presented with the following terms: take a fall on Divad in return for relief on Electric Boat’s cost over runs and clearance to buy the M-1 plant. As it turned out, by the autumn of 1981, Electric Boat had disappeared from the nation’s newspapers, the Navy’s claims had been forgotten, and a further Trident contract had been awarded. Early in 1982, General Dynamics bought the M-1 plant. Its Divad prototype was wheeled into a warehouse.

Ford was not quite home free. In May of 1981, it was chosen as the “source” for Divad, but it did not receive a production contract. Instead Ford was given $159 million and one year in which to correct the twenty-nine “deficiencies” and twelve “shortcomings” that the Army found in its prototype. Then a final production decision would be made.

To judge from Ford’s source contract, the company faced a formidable amount of work. Divad was three tons overweight; 92 percent of its magazine space did not function, so that only a small number of its rounds could actually be fired; its radar often went haywire when tracking tree-top-level helicopters, the very threat Divad had been “optimized” to face; and the computer-monitoring bites had failed to meet even the highly lenient specification of 40 percent reliability. In sum, the contract requested that “accuracy shall be improved.”

Ford quickly moved to consolidate its position. It hired Col. James Crosby, who had been director of the DT/OT tests; he had retired from the Army just after the Ford “source” award was made. Then Westinghouse, the Divad radar supplier and Ford’s largest subcontractor, hired Col. Gary Mahan, a top Fort Bliss official, who had also retired immediately after the source award. Fortune smiled on Ford in September of 1981, when Reagan nominated James Ambrose to be undersecretary of the Army. Ambrose’s last job had been vice president for technical affairs at Ford Aerospace, where he oversaw development of the Divad project.

Late in 1981, Ford’s gun was subjected to a “check test” that was supposed to show that its deficiencies and short-comings had been corrected. The Army will say only that it was satisfied with the check test; the details have not been released to the public. On February 4, 1982, a number of U.S. and British officers assembled at Fort Bliss for a demonstration of the no-longer-deficient Divad. The target was a drone. Huey helicopter, which would hover, motionless. Divad’s electronic brain was switched on. According to one of those present, the gun immediately swung at full speed away from the target and toward the reviewing stand, where the officers were sitting. Brass flashed as the officers dove for cover. Then the gun slammed to a stop, but only because an interlock had been installed the night before to prevent it from pointing directly at the stands. For a while, technicians combed over the machine. Then the Huey drone rose again, and Divad was again switched on. This time it pointed in the direction of the target but began blasting away at the ground just 300 yards out. For the rest of the day, Divad would do nothing but fire into the tumbleweeds. The Huey floated peacefully.

Robert Lyons, Ford’s program manager for Divad, attributes the gun’s behavior that day to water. He explained recently that because the demonstration was for VIPs, Divad was taken to the motor pool the previous evening and washed down, and its electronics were fouled. “Washing it as sheer stupidity,” Lyons said. “The next day, after it dried out, it did fine.” I asked him whether it ever rains in Europe, and he changed the subject.

In May, the Defense Systems Acquisition Review Council (DSARC), the board that makes final production decisions, met to consider Divad. The highest-ranking Army official present was James Ambrose. Several officials representing various Pentagon agencies gave stern critiques of Divad and recommended that it be canceled. “Normally the DSARC is a rubber stamp, but this one was bloody,” a Pentagon source says; “it was the bloodiest one in years.” Ambrose is said to have lost control of his temper and to have risen twice to walk out of the room, declaring that he would not listen to any criticism of his project. “If I’d known you people were going to talk this way, I wouldn’t have come,” Ambrose is reported to have said. (One purpose of a DSARC meeting is to hear objections.) After the meeting, Undersecretary of Defense Richard DeLauer recommended that Divad production be approved. On May 27, Frank Carlucci awarded the contract to Ford.

The contract calls for fifty Divads at $6.3 million each, and an option for 226 more; inflation is expected to bring the full price to $6.8 million. That is only the “target” price. The contract provides for a 25 percent cost overrun with a concurrent incremental increase in Ford’s profit, which could escalate Divad’s price to $8.5 million each. Further overruns are permitted, but beyond 25 percent, Ford will not be paid any extra profit—“a real incentive to stop the overruns there,” Lyons told me. The Army plans to buy 618 Divads; after the first group is built, the price will be renegotiated upward.

Included in the price is ammunition—all Divad will ever need, even in the event of war, the Army says. According to Ford, 24 percent of the Divad purchase price is for ammunition. The “target” price for each proximity-fused round is $275, so the Army will be buying about 6,000 rounds per vehicle. Since Divad holds 502 rounds, that will be enough to load it twelve times. Either the Army has devised a war plan in which its guns are never fired or Divad’s true cost is being systematically disguised until the program is so far along that Congress considers it too late to change anything.

In fact, the Army has devised a plan in which Divad is never fired, at least for training. Because of the expense of the ammunition, the Army plans to hold live firings to a bare minimum and train soldiers on simulators instead. For the Army, this will have the welcome effect of preventing any further graphic demonstrations of Divad’s ineffectiveness, since the gun will rarely be called on actually to knock something out of the sky. But if there should be a sudden war, Divad gunners will go into battle never having fired their weapons.

Divad’s $275 ammunition price compares with a price of $20 each for the 30mm rounds for the cannon on the A-10 airplane. Rounds for the A-10 are about one third the weight of the Divad rounds, but in other respects they are quite similar. There is a considerable difference, though, in the way they are made.

Divad’s ammunition work was awarded on a sole-source basis through Ford to Day & Zimmermann, which operates a GOCO arsenal. Both contractors are being paid cost plus profits, and so have little incentive to hold prices down. At the arsenal, Divad’s rounds will be assembled according to complex “process” specifications that dictate the details of nearly every manufacturing step, drastically reducing efficiency. In contrast, the A-10’s rounds are made by two companies, Honeywell and Aerojet General, which bid against each other for every lot ordered. The low bidder gets the bulk of the order and enjoys the bulk of the profits, but the high bidder gets a portion so that it can stay in the A-10 ammunition business and continue to provide competition. At their plants, Honeywell and Aerojet General assemble A-10 rounds according to simple “performance” standards; so long as the rounds work, the buyer doesn’t care how they’re made. This leaves the contractors free to innovate and find more efficient production methods.

The Army has recently acquired control of the A-10 rounds program from the Air Force, and is taking steps to convert it to operation along Divad lines. The Army’s official reason is that it wants to standardize production.

One reason distorted projects can emerge from a system whose individual members are intelligent and generally well-intentioned is that bureaucratic structure of the Pentagon itself. Multiple independent commands, which compete for attention and budget money, are encouraged to see themselves as the center of all things and their responsibilities to the larger scheme as a hindrance. For example, an original reason for a new anti-aircraft gun was that cannons, unlike missiles, are useful in ground combat as well as against aircraft. Divad, armed with the proper round, could destroy Soviet tanks. Indeed, many military scientists consider an anti-aircraft gun better than a tank in some situations, because it does not need to be aimed precisely and can spray an entire enemy position. Divad, when conceived, was planned for this “dual role.” Since then, the Air Defense Command hierarchy has agitated to have Divad preserved solely for anti-aircraft use. “We wouldn’t want to risk this asset” by having it fire at ground targets, Lieutenant General Maloney says.

In like fashion, Air Defense gets little help in solving its problems from other branches of the Army. One of the Army’s major new projects is the M-2 Bradley “fighting vehicle,” a mini-tank whose armament, a rapid-fire 25mm cannon, would be effective against aircraft. But the Bradley was designed so that its cannon could be elevated only 20 degrees, making it impossible to aim at the sky; anti-aircraft, the designers said, is somebody else’s job. Adding elevation to the Bradley gun, a no-cost, no-problem proposition, was accomplished only on direct orders from the office of the secretary of defense.

The compartmentalization of the Pentagon leads to other crossed purposes. For instance, while Divad’s radar may suit the Air Defense Command’s fascination with “ultra” technology, it makes Divad a dangerous liability from the point of view of field commanders, who are scarcely consulted in the planning of new equipment. Radars broadcast a powerful signal that is easily sensed by cheap, simple radar detectors. Any Divad with its radars on would immediately reveal the division’s location, inviting attack by enemy troops. Equally chilling, radar shines as a beacon to radar-homing weapons. Precision-guided “smart” missiles vary greatly in effectiveness, but one that works with diabolical efficiency is the ARM (“anti-radiation missile”), which locks onto the source of radar emissions. The Israelis recently used ARMs to wipe out Syrian anti-aircraft missiles during the Lebanon war. The missiles have radar control similar to Divad’s.

For this reason, field commanders not only may recommend that Divad’s radars be switched off; they may insist on it—assuming, of course, that the radars are working in the first place. In five years of service on the F-16 fighter, the radar Divad sues has compiled a “mean time between failure” of thirty-four hours. On Divad, the radar will be lucky to match this modest figure. As Divad bumps around in the mud and smoke, it will expose its delicate radars to stresses aircraft radars never encounter. “The Army environment is the toughest environment there is for electronics, much tougher than the air,” says Donald Srull, a scientist who works for the Pentagon’s quasi-federal Logistics Management Institute.

Divad is laid out in “black box” fashion, so that when an electronic component fails (assuming the 40-perent-efficient bits can pinpoint the right part), the entire “box” of circuits can be removed and replaced by unskilled mechanics. But as a result, Divad will require an exceptionally long “tail” of spare parts following it everywhere it goes (including spare parts for its air-conditioners), and keeping up with the demand for parts will be costly and complicated. That problem is no concern of Divad’s designers, however; another department will have to deal with it.

In tacit recognition of Divad’s state, the army has already solicited proposals for a new anti-aircraft cannon, which is known around the Pentagon as “Little Divad” but whose official name is Lads (for “light air-defense”). Lads will be a smaller, simpler system than Divad, probably towed behind a truck like an artillery piece. Pentagon sources believe that Ford has the inside track on Lads. Unfortunately, they say, Ford’s proposal repeats most of Divad’s flaws. The existence of the Lads program, however, could be converted into an opportunity to provide the Army with a realistic, affordable anti-aircraft weapon. First, the Divad program might be canceled or at least terminated at the end of the fifty-unit contract already signed. Then Lads could be designed to combine inexpensive shells with simple visual aiming aids.

General Dynamics is working on a prototype of a 35mm anti-aircraft gun on an armored chassis much smaller and lighter than Divad’s. Instead of radars, the gun would have a laser range-finder (lasers of this type are relatively inexpensive) and a Flir for operation on clear nights. According to industry sources, the company is preparing this weapon solely for the export market, because it assumes that the U.S. Army is no longer interested in anything short of “ultra” technology. It should, however, be well worth congressional scrutiny.

General Electric’s long-discarded Divad proposal is also worth a second look. The company’s 30mm cannon (which is built for the A-10 aircraft) was disqualified because its rounds are ineffective at four kilometers. But at shorter distances, they would be more than adequate.

General Electric’s gun, even with radars and computers, was by far the cheapest Divad contender, priced at about $1.1 million in 1977. With a laser-and-Flir set instead of electronic fire-controls it should cost about the same today. The 30mm cannon, meanwhile, though not NATO-interoperable, offers an extremely high rate of fire with the cheap $20 shell. In one minute, it can fire seven times as many shells as Divad, for half the price.

A cheaper, simpler gun would allow the Army to cut defense spending while building a larger number of weapons—enough to have some impact on a future war, or, so much more important, to help prevent one. Buying just 618 Divads translates into only one new gun per tank column—hardly enough to deter a potential enemy.

If the Divad program is allowed to continue its present course, even that one gun may be absent from the field. Because of the extraordinary price and delicacy of the Divad “asset,” the latest tactical plans, according to Pentagon sources, call for tanks to “secure” a battle area first and then call up Divad from behind. After $5 billion is spent, contractors and bureaucrats will be happy, but the Army will be right back where it started—without an air defense for soldiers dying in battle.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.