Precision Strike

In May 1965, Colonel Joe Davis Jr., commanding the Air Force’s Armament Laboratory at Eglin Air Force Base in Florida, walked into a conference room and asked a simple question. American pilots had been flying over North Vietnam for months, and the gun camera footage was humiliating. Bombs missed bridges by hundreds of feet. Whole squadrons were disappearing into anti-aircraft fire to drop tonnage that vanished into rice paddies. Davis wanted somebody, anybody, to figure out a way to make a bomb hit what the pilot was actually aiming at.

A few months later, an engineer at Texas Instruments named Weldon Word answered him.

Word ran a small group inside TI’s defense electronics business in Dallas. He was a practical man, more interested in getting hardware to work than in the elegant solutions favored by the bigger contractors. When the Air Force put out a call for guided-bomb concepts, the major defense primes pitched gold-plated proposals built around expensive new airframes. Word’s pitch was almost insultingly simple. Take a regular dumb bomb, the kind already stacked by the thousands in every Air Force depot, and bolt a guidance kit onto its nose and tail. The kit would carry a laser seeker on the front and four movable canard fins. A second airplane, or the bomber itself, would shine a laser on the target. The bomb would steer itself down the reflected beam.

The trick was the seeker. Word’s team, working with TI’s solid-state engineers, used a quadrant photodetector, a flat sensor split into four pie-slice segments. Reflected laser light would land on the detector, and a small analog circuit would compare the signal strengths from each quadrant. If the upper-left segment received more light than the others, the bomb was drifting low and to the right; the canards twitched to correct. Round and round, dozens of times a second, all the way down. The whole guidance assembly was small enough to mount in a nose cone the diameter of a coffee can. Inside it sat early integrated circuits of the kind TI had been making for the Apollo program and the Minuteman missile, repurposed for a job no one had quite figured out how to do.

Texas Instruments named the program Paveway. The contract, awarded in 1967, was modest by Pentagon standards: a few million dollars to deliver a working prototype kit. By internal TI accounts later collected by IEEE Spectrum and the Smithsonian’s Air and Space Museum, Word’s team built early seekers by hand on benches in Dallas, soldering components and testing in the parking lot. The first guided drops at the Eglin range in 1967 went well enough that the Air Force ordered the kits into combat use in 1968. The first laser-guided bombs, designated KMU-342 and built around the standard 750-pound M117 bomb body, were carried into North Vietnam by F-4 Phantoms of the 8th Tactical Fighter Wing, the Wolfpack, flying out of Ubon Royal Thai Air Force Base under Colonel Robin Olds and his successors.

Almost no one outside the squadrons noticed.

That was deliberate. The Air Force kept the program quiet because it was not yet clear how reliably the kits would work in combat. Monsoon clouds blocked laser energy. The aircraft designating the target had to fly a steady orbit, exposed to anti-aircraft guns. Smoke from earlier bombs scattered the beam. In the wrong weather, a Paveway flew dumb. But on a clear day, against a target painted by a competent weapons systems officer, the bombs were doing something nobody had seen before. They were hitting the aiming point. Not within a hundred feet, not within fifty. The aiming point.

To understand how strange this was in 1968, it helps to remember what bombing accuracy looked like before. In World War II, the Eighth Air Force’s daylight precision campaign over Germany had a circular error probable, the radius around the target inside which half the bombs fell, that planners measured in thousands of feet at high altitude. By Korea and the early years of Vietnam, with better optical sights and radar bombing, CEP from medium altitude had improved to a few hundred feet on a good day. Air Force studies of Rolling Thunder, the air campaign over North Vietnam from 1965 to 1968, found that even experienced crews dropping iron bombs from level flight typically achieved CEPs around 400 feet against fixed targets. Hitting a single bridge span required dozens of sorties, scores of bombs, and the willingness to lose airplanes and crews in the process.

The Paveway, in early operational use, was producing CEPs in the range of 20 to 25 feet. The Air Force’s own assessments after 1968, later compiled in Air University and Maxwell AFB historical studies, found that roughly half of laser-guided bombs were direct hits, defined as within a few feet of the aim point. The rest landed close enough that the difference was, for most targets, immaterial. A seventeenfold improvement in accuracy at a kit cost in the low single-digit thousands of dollars per bomb was not an incremental advance. It was a change of kind.

Nowhere did this become clearer than at a steel-and-concrete bridge eighty miles south of Hanoi, on the road and rail line connecting North Vietnam to the south.

The Thanh Hoa Bridge, called Ham Rong by the Vietnamese and Dragon’s Jaw by the Americans, crossed the Song Ma River in two long spans on a single massive concrete pier. It carried Highway 1 and the rail line over the river. Cutting it would force the People’s Army to detour every truck and rail car bound for the front lines through ferries and rough secondary roads. The Pentagon’s planners called it one of the most lucrative interdiction targets in the country.

The Air Force and Navy started bombing it in April 1965. They kept bombing it for the next seven years.

The first raid, on April 3, 1965, sent forty-six F-105 Thunderchiefs against the bridge with 750-pound bombs and Bullpup air-to-ground missiles. The Bullpups, radio-controlled and steered by a pilot watching a flare on the missile’s tail, bounced off the bridge’s concrete pier like marbles thrown at a wall. The F-105s came back the next day with a heavier load and lost two aircraft to anti-aircraft fire. By post-strike reconnaissance, they had inflicted no significant damage.

This pattern repeated. Strike packages of twenty, thirty, forty aircraft would fight their way through MiG patrols and the densest surface-to-air missile defense the world had ever seen. They would drop hundreds of bombs at a structure designed by French and Vietnamese engineers to carry rail loads. They would lose airplanes. The bridge would still stand. By the count later assembled in Air Force historical reports, more than 800 sorties had been flown against the Dragon’s Jaw between 1965 and 1968. Eleven American aircraft were lost over and around the bridge. The structure was scarred, pitted, blackened by fire. It still carried trucks and trains.

In May 1967, the Navy tried something new. A-4 Skyhawks from the carrier Bon Homme Richard launched the AGM-62 Walleye, a glide bomb developed at the Naval Weapons Center at China Lake and built by Martin Marietta. The Walleye was a different bet from the Paveway. Instead of laser guidance, it carried a small television camera in its nose. The pilot, before release, would slew crosshairs in his cockpit display over a high-contrast feature on the target, lock the camera onto the contrast pattern, and pickle the bomb. The Walleye would then glide toward whatever its camera was tracking, refining its aim using a pattern-matching circuit that compared each frame to the locked image.

It worked, after a fashion. Several Walleyes hit the Dragon’s Jaw on those May 1967 raids. They left holes in the bridge deck. They did not drop a span. The Walleye carried a relatively small warhead, around 825 pounds, and the bridge’s structural members absorbed the punishment. Naval aviators called it a precision weapon attached to too small a bomb.

Then came 1972.

In late March of that year, the People’s Army of Vietnam launched the Easter Offensive, pouring conventional forces across the Demilitarized Zone in the largest assault of the war. President Nixon, looking for a way to halt the offensive without putting more American troops on the ground, ordered the resumption of bombing over the North under a new operation, Linebacker. This time the rules of engagement were looser, the target list was longer, and the Air Force and Navy had stockpiled thousands of laser-guided bomb kits.

On April 27, 1972, a flight of F-4 Phantoms from the 8th Tactical Fighter Wing, the same Wolfpack that had carried the first Paveways into combat four years earlier, attacked the Dragon’s Jaw with a mix of laser-guided and electro-optically guided bombs. According to USAF historical office accounts and later analysis by Marshall Michel, the strike damaged the western span and dropped it onto its abutment but failed to fully sever the bridge. North Vietnamese engineers, expert by now at battlefield repair, began patching it.

On May 13, 1972, fourteen F-4s went back. Eight of the aircraft carried 2,000-pound and 3,000-pound laser-guided bombs. The remainder flew escort and laser designation. The strike leader was Captain D.L. Smith. Over the bridge, the designators painted the steel at the heart of the structure with an invisible beam from above. The bombers released. The bombs flew themselves into the bridge.

When the smoke cleared, the western span of the Dragon’s Jaw was in the river. Reconnaissance photos taken in the following days, declassified later and published in Air Force historical surveys, showed the structure broken in a way no amount of unguided bombing had managed in seven years. The North Vietnamese began rerouting traffic around the wreckage that night.

William Beecher’s dispatch in The New York Times that month, datelined Saigon, described the strike soberly. American officials, Beecher wrote, were claiming that “smart bombs” had finally accomplished what hundreds of conventional sorties had failed to do. The Pentagon released grainy black-and-white footage of bombs corkscrewing down toward bridges and warehouses. To viewers used to the messy ambiguity of Vietnam, the imagery felt almost staged. It was not. It was simply what guided weapons looked like.

Across Linebacker, the Seventh Air Force later reported that laser-guided bombs achieved direct-hit rates around fifty percent against fixed targets, an order of magnitude improvement over unguided ordnance. Single airplanes were now destroying bridges that had previously consumed wings of aircraft and weeks of effort. The kill ratio per sortie inverted. Mark Clodfelter, in his study The Limits of Air Power, would later argue that Linebacker’s tactical successes did not by themselves win the war. Political constraints and North Vietnamese resilience set the terms of the conflict more than airpower did. But the technical demonstration was unambiguous. Precision had arrived.

What is easy to miss, looking back at the gun camera footage, is that the revolution was not really about the bomb. It was about the chip.

The Paveway’s seeker head, the small electronics package that read the laser signal off the photodetector and translated it into canard commands, was made of integrated circuits. By the mid-1960s, TI was producing silicon ICs in volume for the Apollo Guidance Computer and for the Minuteman II’s guidance system. The same logic gates and op-amps that had flown men to the Moon and held nuclear deterrence aloft were being repackaged, in cheaper and more rugged form, into the nose of a $3,000 bomb kit. Without the IC, the Paveway’s seeker would have required vacuum tubes or discrete transistors filling a much larger volume than a bomb’s nose section could spare, drawing more power than its battery could provide, and failing far more often than its mission demanded. Solid-state electronics, miniaturized onto silicon, were the precondition for the whole concept.

The same was true of the Walleye. Its pattern-matching circuit, the heart of the early TV-guided bomb, depended on integrated logic to compare frame to frame in real time. Earlier engineers had imagined television-guided weapons as far back as the Second World War, when the Army Air Forces had experimented with a camera-equipped GB-4 glide bomb and found it almost unusable. The technology had been waiting for chips small enough and reliable enough to do the math. By 1967 they existed.

For the Pentagon, the implication was strategic and immediate. If one bomb could destroy what had previously required a hundred, then a smaller force could do the work of a larger one. A bridge dropped by a single F-4 was a bridge that did not have to be dropped by a flight of B-52s. A tank killed by a single guided artillery round was a tank that did not have to be killed by a battalion of friendly armor. The arithmetic of conventional warfare, set since the first machine gun, began to bend.

It also implied something about the relationship between weapons and the industries that built them. For the first time, the decisive ingredient in a weapon’s effectiveness was not its explosive payload or its airframe performance or even the courage of its crew. It was the quality of the silicon riding inside its guidance package. A nation that could not manufacture good integrated circuits could not, at the limit, build effective modern weapons.

In Dallas, Weldon Word’s group at Texas Instruments noticed none of this geopolitical weight at the time. They were busy. The Air Force kept ordering more Paveway kits. The Navy was experimenting with laser variants of its own weapons. Other contractors were lining up to build seekers, designators, and follow-on kits. Paveway I would, by the early 1980s, evolve into Paveway II with proportional guidance, then Paveway III with imaging infrared, each generation depending on more capable, more compact integrated electronics that the commercial chip industry happened to be making available on the Moore’s Law curve. Word’s small team had pulled a thread that would unravel a century of bombing doctrine. They were mostly just glad the prototype worked.

The Soviet General Staff studied the news from Vietnam with care. They understood, sooner than many in Washington did, that the ground had shifted, and that the integrated circuits underlying the new American weapons were not something the Soviet bloc could easily match. Their own chip manufacturing, centered at Zelenograd, was already falling visibly behind.

The line from Weldon Word’s bench in Dallas would, twenty years later, run through the open desert of Iraq. But it ran first through a single bridge in the Song Ma valley, dropped in May 1972 by bombs that knew where they were going.

In the interim, the chip industry that had made it all possible was about to do something that would shape the rest of the century. It was about to leave home.