War Hero

At two thirty-eight in the morning on January 17, 1991, Major Greg Feest was alone in the cockpit of an F-117A Nighthawk, holding a steady track over the western Iraqi desert at twenty-five thousand feet. He was carrying two two-thousand-pound laser-guided bombs and the responsibility, although nobody had said it in those words, for opening the war.

The Nighthawk did not look like a fighter. Its facets were arranged for radar-cross-section, not for lift, and it flew like a brick when the autopilot was off. What it did have was a small infrared targeting turret in its belly and, in its avionics bay, a stack of computer cards built around commercial integrated circuits. The cards ran the flight controls, drove the navigation, and aimed the bomb. Feest’s job, on the way in, was mostly to monitor.

Forty-two minutes earlier, helicopters from Task Force Normandy, eight Apaches and four Pave Lows, had skimmed across the border at low level and obliterated two Iraqi early-warning radars near the Saudi line. That was the door. Feest was the man walking through it. His target was an interceptor operations center at Nukhayb, a node in Iraq’s air-defense network, and he had been told by the planners in Riyadh to put both bombs into a single specific room. He centered the crosshairs in his targeting screen on the marked entry point, held the laser on it, and released. The first bomb went through the roof and detonated. He came around again, painted a second aim point, dropped the second weapon. Then he turned south and went home, untouched. By the time he landed at Khamis Mushait on the southern edge of Saudi Arabia, the sky over Baghdad was already on television.

It was on television because a thirty-five-year-old satellite uplink, two reporters, a New Zealand–born war correspondent who had stayed in country against his network’s wishes, and a CNN field producer named Robert Wiener had spent the previous weeks negotiating, pleading, and bribing their way into a four-wire phone circuit out of room 906 of the Al-Rasheed Hotel. When the bombs began to fall a few miles from their window, Bernard Shaw, John Holliman, and Peter Arnett were the only Western correspondents broadcasting live from inside the Iraqi capital. CNN’s audio director in Atlanta heard Shaw say into his microphone, “Something is happening outside.” Then Holliman pushed the heavy black handset of the four-wire close to the open window so that fifty million people watching at home in the United States, and millions more in Europe and Asia following the feed, could hear the antiaircraft fire and the dull thumps of distant impacts in real time. Shaw’s voice, when he came back on, was almost apologetic. “This feels like the center of hell,” he said.

That sentence, broadcast at 6:35 p.m. Eastern, scooped the Pentagon’s official announcement of hostilities by twenty-seven minutes. It also performed a different and more permanent function. Up to that night, the public image of war had been edited. From World War II newsreels through Vietnam gun-camera footage, what civilians saw was always after the fact, framed by a producer, narrated by a man who had read the script. The Baghdad broadcast was the first time war had played live on a continuous global feed, with the noises in the right channel and the lights in the right window, while it was still unclear who was winning and whether the people in the hotel would survive the night. The medium had finally caught up to the technology that was producing the spectacle.

What that technology was, exactly, took a few weeks for the world to absorb. The Iraqi army standing in the way of the U.S.-led coalition was, by any conventional measure, formidable. Britannica, the Council on Foreign Relations, and Pentagon planners alike described it as the fourth largest in the world, around 900,000 men under arms, more than 5,000 tanks, dug into prepared positions in Kuwait and southern Iraq behind triple belts of mines and berms. It had eight years of recent combat experience from the war with Iran. It was equipped with Soviet T-72s, MiG-29 fighters, and a French and Soviet integrated air-defense system that planners in Washington considered, in some respects, denser than what the Soviets fielded around Moscow itself. The American intelligence consensus had been that any ground assault would cost the coalition somewhere between five and twenty thousand casualties.

The Black Hole disagreed.

The Black Hole was the basement of the Royal Saudi Air Force headquarters in Riyadh, where Brigadier General Buster Glosson and a small team of Air Force planners worked under Lieutenant General Charles Horner, the air commander, on a daily air tasking order that ran to thousands of pages. The room had been nicknamed for its windowless secrecy and the way its occupants disappeared inside it for days at a time. What Glosson and Horner were drafting was an air campaign of a kind that had never been executed. They had been handed, by the recent generations of Pentagon investment in stealth, in laser and electro-optical seekers, in cruise missiles, and in space-based reconnaissance and navigation, a set of weapons whose collective effect they were the first to plan around. Their war was not Vietnam at higher tonnage. It was something else.

The opening salvos showed it. While Feest’s two bombs were still in the air over Nukhayb, the cruisers San Jacinto and Bunker Hill were launching the first of more than a hundred Tomahawk land-attack missiles fired in the first twenty-four hours of the war. The Tomahawk, a twenty-foot turbojet missile that flew at the height of telephone poles and used terrain-comparison radar and an optical scene-matching camera to hit aim points in the centers of cities, had been a Cold War weapon designed to penetrate Soviet air defense from beyond range. It was being used here as a long-range surgical instrument, programmed to thread streets in central Baghdad and hit specific buildings without putting a pilot in the air. By the war’s end, U.S. Navy ships and submarines had launched 288 Tomahawks; over the full conflict, more than 300 Block II variants flew, and Navy after-action analysis put their successful arrival rate above eighty-five percent. They appeared on the CNN feed as silent shapes drifting between Baghdad’s high-rises, occasionally illuminated for a half-second by tracer fire that never quite caught up to them.

The B-52s contributed cruise missiles too, AGM-86 launches from aircraft that had taken off from Barksdale Air Force Base in Louisiana, flown across the Atlantic, fired their weapons over the Mediterranean, and returned to Louisiana, a thirty-five-hour round trip ending at the same hangar, with the entire combat portion executed without crossing into Iraqi airspace. The geometry of distance had collapsed.

Inside Iraq, by sunrise on January 17, the air-defense system that had looked formidable on paper was already stunned and partly blinded. The F-117s, which represented two and a half percent of the coalition’s fixed-wing combat aircraft, hit thirty-one percent of the targets struck on opening night. The stealth fighters’ radar-defeating geometry, paired with their laser-guided bombs, allowed them to fly straight at command bunkers, telephone exchanges, and SAM headquarters that earlier generations of strike packages would have approached only with elaborate suppression escorts and high losses. Wave after wave came in. By the time Saddam Hussein’s air defense could begin to reorient, the integrated network on which it depended, the radars and the command centers and the fiber-optic exchanges, was already coming apart.

A Pentagon briefer, asked a few days into the war how the Air Force had achieved this, gestured at a video and let the picture answer.

The picture, by the second week of the war, had become a genre. General Norman Schwarzkopf, the theater commander, took to opening his Riyadh briefings with cockpit footage from F-117s and F-15Es. The format was simple. Black-and-white infrared video, the cross-hairs of the targeting pod centered on a building or a bridge, the calm voice of the weapons systems officer counting down, then the blossom of light that filled the screen as the laser-guided bomb arrived. On January 30, Schwarzkopf showed a clip taken over a bridge in central Iraq. A small white shape, a truck, moved across the bridge from left to right. The cross-hairs tracked the bridge’s central span, not the truck. A bomb came in from the upper right, the bridge collapsed, and the truck cleared the far end with what looked like a meter to spare. “And now in his rear-view mirror,” Schwarzkopf said, “the luckiest man in Iraq.” The room laughed. The clip ran for years afterward on cable news, on military recruiting reels, on documentary specials. It was the single most watched piece of bombing footage ever produced.

What it concealed was that most of the bombs being dropped on Iraq were not like that.

The Gulf War Air Power Survey, a multi-volume study commissioned by the Air Force after the war and led by the historian Eliot Cohen, eventually put a number on it. Of all air-delivered ordnance dropped on Iraqi targets between January 17 and February 28, 1991, only about eight percent was precision-guided. The other ninety-two percent consisted of the same kind of unguided iron bombs that had been dropped over Germany in the 1940s, over Korea in the 1950s, and over Vietnam in the 1960s, dropped in this case largely from B-52s flying high-altitude carpet patterns over Iraqi divisions in the field. The headline image of the war was the cross-haired bridge. The actual tonnage looked a lot more like Linebacker. General Merrill McPeak, the Air Force chief of staff, would later concede that precision weapons had constituted barely seven percent of the total bomb tonnage.

But the surveys also showed who had done what. The eight percent of precision-guided weapons accounted for the great majority of strategic-target damage. Command bunkers destroyed, bridges dropped, hardened aircraft shelters opened up, oil refineries shut down. The dumb-bomb ninety-two percent had been used overwhelmingly against deployed Iraqi divisions in the open desert, a job for which carpet-bombing remained, brutally, well suited. The two halves of the campaign were complementary. The smart bombs collapsed Iraq’s nervous system; the dumb bombs broke its body. Both were necessary; only one made television.

The same pattern showed in the air. Fighter pilots flying F-15s and F/A-18s ran intercepts that depended on a chain of Boeing E-3 Sentry AWACS aircraft cruising in long racetracks behind the lines, their rotodomes painting the sky over Iraq with a continuous radar picture relayed by encrypted datalink to the strike packages below. On the ground side, two Boeing E-8 Joint STARS aircraft, prototype platforms still officially in development, flew forty-nine combat sorties carrying side-looking radars that could distinguish moving vehicles from static ones across an entire Iraqi corps sector at once. The systems were so new that the test pilots flying them were also their operational pilots. Brigadier General John Stewart, the senior intelligence officer for Army Central Command, would later call Joint STARS “the single most valuable intelligence and targeting collection system in Desert Storm.” A coalition strike pilot launching from Saudi Arabia could, in principle and often in practice, have his target nominated by Joint STARS, his approach corridor cleared by AWACS, his attack run logged by U-2 imagery, and his bomb guided by laser energy from a second aircraft trailing behind him. The killing was distributed across a network. The chips connecting the network were what made the network possible.

GPS made its debut as a battlefield system in the same conflict. Only sixteen Navstar satellites were on orbit by the start of Desert Shield, and military receivers were in painfully short supply. The Army’s primary handheld unit, the AN/PSN-10 Small Lightweight GPS Receiver, known to its users as the Slugger, was a four-pound brick that looked like a 1980s calculator and gave a position fix accurate to about sixteen meters. By the start of the ground war, the Pentagon had managed to push roughly four thousand Sluggers and a handful of vehicle-mounted receivers into theater, supplementing them with commercial Trimble and Magellan units bought on the open market. It was nowhere near enough for every squad, but it was enough for the lead vehicles of the armored columns that, on the night of February 24, began rolling north out of Saudi Arabia into the open desert. The Iraqis had assumed any coalition advance would have to come up the coast, where roads existed. The coalition’s left wing, instead, drove straight across an unmapped wasteland that Iraqi planners had assumed nobody could navigate. The Sluggers told them where they were.

For the units in front, the experience of the ground war was less the precision spectacle of the air briefings than a long, fast hallucination. On February 26, in the middle of a sandstorm so dense that visibility shrank below three hundred meters, Captain H.R. McMaster’s nine M1A1 Abrams tanks of Eagle Troop, 2nd Armored Cavalry Regiment, came across an Iraqi Republican Guard armored brigade dug in along a north-south map line called the 73 Easting. The Iraqi T-72s could not see the Americans because the sandstorm blinded their optical sights. The American thermal imagers, looking for the hot signatures of running engines and warm steel hulls, did not care about visibility. McMaster’s troop opened fire at over two thousand meters and destroyed twenty-eight tanks, sixteen armored personnel carriers, and roughly thirty trucks in twenty-three minutes. Eagle Troop took no casualties. The wider Battle of 73 Easting cost the U.S. 2nd Armored Cavalry Regiment one M3 Bradley and around a dozen wounded; on the Iraqi side it left, by post-action survey, more than 160 destroyed armored vehicles. It was the largest tank engagement involving American forces since the Second World War, and it ended without the losing side having seen, in any meaningful sense, who hit them.

That story repeated itself across the front. The Abrams’s gunner had a thermal sight derived from the same family of cooled mercury-cadmium-telluride detectors, packaged behind the same kind of read-out integrated circuits, that the F-117 had carried in its targeting pod. The fire-control computer that took the gunner’s lased range, the wind reading, the cant of the hull, and the ballistic profile of the round, and adjusted the gun’s elevation and lead in a fraction of a second, was a small armored box of digital electronics that the original Abrams designers in the 1970s had built around the still-young microprocessor industry. The Iraqi T-72 had a coincidence rangefinder that worked when the gunner’s eye was clear and aimed, and not very well when it was not. By the time of the Gulf War, two decades of divergent investment in semiconductors had concentrated themselves in the gap between those two fire-control systems. The result was that hundreds of Iraqi crews died without firing back.

The coalition lost, in total, 31 tanks across the entire ground campaign. The Iraqi army lost more than 3,000. Coalition combat deaths came to roughly 240 over six weeks of war, against Iraqi battle dead estimated at the time in the tens of thousands. President George H. W. Bush, watching the imbalance grow on the morning of February 28, ordered a halt at the hundred-hour mark of the ground campaign. The decision was political, partly humanitarian, partly a reluctance to be seen on television destroying a fleeing army.

That image had already been produced. On February 26 and 27, a column of Iraqi vehicles, military and civilian mixed together, fled north from Kuwait City along Highway 80 toward Basra. Coalition aircraft caught them at the Mutla Ridge pass, where the road bottlenecked. A-6 Intruders dropped cluster munitions on the head and tail of the column, freezing it in place. Then, for hours, Air Force and Marine and Navy aircraft worked the wreckage. By the time the last sortie ran, an estimated 1,400 to 2,000 vehicles, many of them stolen Kuwaiti civilian cars driven by Iraqi conscripts and looters, were burning in a stalled line stretching for miles. The road was renamed by the press, almost immediately, the Highway of Death. The footage that came back was harder to laugh at than Schwarzkopf’s truck on the bridge. Bush’s decision to stop, made over the next thirty-six hours, was driven in part by aides who told him that another day of those images would erase the political capital the war had banked.

Underneath the spectacle, the war produced its own gritty footnotes. The most uncomfortable of them was the Patriot.

The MIM-104 Patriot was a Raytheon-built surface-to-air missile, originally designed to shoot down Soviet aircraft, hurriedly upgraded in the 1980s with a software package called PAC-2 that allowed it to engage tactical ballistic missiles. When Iraq began firing Scud missiles at Saudi Arabia and at Israel on the second night of the war, the Patriot was the only missile defense the coalition had. The Bush administration deployed Patriot batteries to Riyadh, Dhahran, and, in an unprecedented gesture, to Tel Aviv and Haifa, in part to keep Israel from joining the war and shattering the Arab coalition. Television coverage that month featured a near-nightly ritual: a streak of orange light rising into the sky over the desert, an explosion at altitude, applause in the press tent. The Pentagon told reporters that the Patriot was hitting close to nine in ten Scuds over Saudi Arabia and roughly half of those over Israel. Raytheon’s stock surged. Schwarzkopf called the Patriot batteries “magnificent.”

The reality, as later assessments by the General Accounting Office, MIT physicist Theodore Postol, and Israeli air-defense officials made clear, was considerably murkier. Reviewing radar data, debris recovery, and ground damage reports, GAO analysts in 1992 concluded that only about nine percent of Patriot-Scud engagements during the war could be supported by strong evidence of an actual warhead kill. Israeli studies were even harsher; Defense Minister Moshe Arens and General Dan Shomron later said the Patriot had achieved at most one or two confirmed kills over Israel. What had often happened on television was that a Patriot fuse had triggered close to a Scud’s airframe but not its warhead, scattering debris that caused as much ground damage as the original missile would have. The political function of the Patriot, in keeping Israel out of the war and reassuring Saudi cities, had been real. The advertised technical performance had not. The chip-enabled weapons of the Gulf War were, in this one prominent case, not nearly as smart as the briefings suggested.

The most sober reminder of that came on February 25, in Dhahran. A barracks housing the Pennsylvania-based 14th Quartermaster Detachment, a reserve water-purification unit, was hit by a Scud. Twenty-eight American soldiers were killed and roughly ninety-eight wounded, the worst single American loss of the war. The Patriot battery that should have engaged the incoming Scud had not fired. Investigators traced the failure to a software bug in the system’s range-gate timing. The Patriot computer kept time in a twenty-four-bit fixed-point register; it multiplied tenths of a second by an approximation of one tenth that, because one tenth has no exact binary representation, drifted by a tiny amount each cycle. After the battery had been running continuously for around a hundred hours, the cumulative error had grown to about a third of a second. A Scud, traveling at roughly a mile per second, was therefore searched for in the wrong patch of sky. Raytheon had identified the bug and shipped the patched software days earlier. The patched tape arrived in Dhahran the day after the strike.

The story made its way into computer-science textbooks within a few years. It made its way more slowly into the public consciousness. The Patriot would survive the war as a political symbol, sold to allied capitals from Seoul to Tel Aviv as the cornerstone of late-twentieth-century missile defense, and it would be improved across successive software builds until it became something close to its original advertising. But the irony of the moment was hard to miss. In a war that had been sold as the triumph of computing power applied to combat, the worst single American casualty count had come from a floating-point arithmetic error.

None of which dimmed the strategic impression of the war on the people most carefully watching.

In Moscow, the Soviet General Staff studied the campaign in detail through the spring of 1991. The work was led by figures like Vladimir Slipchenko, a major general who had spent the previous decade thinking about what Marshal Nikolai Ogarkov had called the reconnaissance-strike complex, the integration of sensors, command networks, and precision weapons that Ogarkov had warned, as early as 1982, would dominate the next war. Slipchenko’s later writing was unsparing. Desert Storm, he argued, was the first concrete demonstration of a sixth-generation war: a war waged at long range by precision weapons cued by space-based and airborne sensors, in which the side without those tools was structurally incapable of winning, regardless of mass. Soviet armor and Soviet doctrine, both bought in volume by Saddam Hussein, had been demolished in plain view by an opponent whose decisive advantage lay in silicon. The Soviet Union itself had less than a year to live, and its successors would spend decades trying, with varying intensity and limited success, to digest what they had seen.

In Tokyo, the war was watched through a different lens. Japanese newspapers carried the cockpit footage and the Schwarzkopf briefings without much commentary about industrial implications. They did not need to. The integrated circuits inside the F-117’s avionics and the Tomahawk’s guidance computer were, in many cases, made on Japanese fabrication equipment, sometimes in Japanese fabs, although the high-end logic and the radiation-hardened parts remained American. Japanese officials who had spent the late 1980s arguing that their country’s chip dominance gave it strategic leverage over the United States had to reckon, as they watched the Iraqi army come apart on television, with what kind of leverage that was. Strategic leverage assumed both sides agreed on what the chips were for. The U.S. government had just demonstrated, on global television, what it thought they were for.

In Washington, the assessment process was less philosophical and more concrete. Andrew Marshall, the durable head of the Pentagon’s Office of Net Assessment, had been writing memos about a coming “military-technical revolution” since the late 1970s, drawing partly on Soviet writing about the same subject. His office spent the months after Desert Storm collecting after-action reports, interviewing pilots and tankers and intelligence officers, and asking what would now be possible at scale. Paul Wolfowitz, the under secretary of defense for policy, and his deputy Scooter Libby drafted a 1992 Defense Planning Guidance that the press, when leaked excerpts emerged, would dub the Wolfowitz Doctrine. Among its less-noticed sections was a quiet observation that smart munitions were no longer experimental, that the United States now possessed a generational lead over any potential adversary in their use, and that maintaining that lead would require sustained investment in the underlying technologies of sensors, communications, computing, and the chip industry that produced them. The doctrine that would come to be called the Second Offset, the idea that the United States would substitute precision and information for mass, ceased after January 1991 to be a theory. It was now a demonstration with footage.

For the chip industry itself, the Gulf War’s most immediate effect was reputational. American semiconductor companies had spent the late 1980s defending themselves against the impression, fed by a decade of Japanese DRAM dominance and the visible wreckage of GCA and other equipment makers, that the United States no longer had a competitive industry. The Gulf War rebutted that impression in a particular and lopsided way. The high-end logic chips inside Texas Instruments’ Paveway seekers, inside the targeting pods at Martin Marietta and Hughes, inside the missile-guidance computers at Raytheon and General Dynamics, inside the navigation receivers at Trimble and Magellan, inside the flight-control systems at Lockheed and Northrop, were almost without exception American silicon. The chips Japan had taken from American firms were memory, the storage cells of the computer industry. The chips that had won the war, the kind of chips on which the Pentagon wrote checks regardless of unit cost, were logic and analog parts where American companies still led. By the end of 1991, that fact had begun to register inside Sumitomo and Toshiba and Hitachi as well as inside Intel and TI. A rebalancing was beginning.

Bernard Shaw left Baghdad on January 18, escorted by Iraqi minders to the Jordanian border. Peter Arnett stayed for the duration, broadcasting under censorship from a country that was being progressively reduced to its eighteenth-century infrastructure by weapons that knew exactly where they were going. CNN’s prime-time ratings during the war averaged roughly ten times its prewar baseline. The expression “smart bombs” entered everyday English. Ron Howard’s later film about the network’s Baghdad coverage was titled “Live from Baghdad,” because that, more than any of the actual reporting, was the news.

What viewers around the world saw, for six weeks beginning at 6:35 p.m. Eastern on January 16, was the future of warfare. They saw, although most of them did not name it, the future of the world that depended on a single industry. The cross-hairs on the bridge, the Tomahawk threading the Baghdad street, the green-on-green thermal silhouette of the Iraqi tank vanishing in a flash, the silent overhead radar tracks of armored columns in the dark, none of those images existed without integrated circuits made to specifications no command economy and few civilian markets had any reason to demand. For thirty years, the Pentagon’s offset bet on chips had been an argument in academic journals and white papers. In January 1991, on a planet’s worth of television sets, it became something else.

It became proof.