The Secret Sentry (16 page)

At six forty-five p.m. EDT, thirty-eight minutes after McNamara had sent the air strike execute order to CINCPAC, President
Johnson met with sixteen senior congressional leaders from both parties and briefed them for ninety minutes, informing them
that he had authorized retaliatory air strikes against North Vietnam and would seek a congressional resolution in support
of his action.
⁷³

But the conflicting reports sent by the
Maddox
and the
Turner Joy
had created consternation at the Pentagon and at CINCPAC, both of which desperately wanted uniform and consistent reports
from both ships as to what had occurred the previous night. Sharp sent a message to Herrick asking, “Can you confirm that
you were attacked by PT or Swatow?” Herrick did not respond to the request, but the captain of the
Turner Joy
radioed at six ten a.m. local time (seven ten p.m. EDT, August 4) that he was convinced that an attack had taken place because
a lookout had reported seeing a torpedo wake.
⁷⁴

The mounting number of conflicting reports from the
Maddox
and the
Turner Joy
only created more concern at higher headquarters. At eight a.m. GOT time, August 5, the commander of the Seventh Fleet, Admiral
Roy Johnson, asked the captain of the
Turner Joy
for the names of the witnesses to the attack and an evaluation as to their reliability. Thirty minutes later, Johnson ordered
the captains of the
Maddox
and the
Turner Joy
to initiate a search for debris that would prove that there had been a battle on the night of the 4th. After a twenty-minute
search, both ships were forced to report that they had found no debris at the alleged site of the sea battle.
⁷⁵

At ten thirty p.m. EDT on August 4, while navy commanders in the Pacific were still furiously trying to collect and collate
the evidence, President Johnson went on television to announce, “Air action is now in execution against gunboats and certain
supporting facilities in North Vietnam.” As he spoke, sixty-four U.S. Navy fighter-bombers from the aircraft carriers
Ticonderoga
and
Constellation
struck North Vietnamesenaval bases, surface units, and oil storage depots, destroying or damaging twenty-five patrol and torpedo
boats and more than 90 percent of North Vietnam’s petroleum storage capacity. The toll for America, however, was heavy. North
Vietnamese antiaircraft gunners shot down two navy fighter-bombers, resulting in the first American prisoner of war (POW)
and the first pilot confirmed dead in the Vietnam War.

In the White House and the Pentagon’s haste to execute the air strikes, nobody bothered to tell NSA that it was happening.
As NSA director Gordon Blake told an interviewer, “the retaliation took everyone by surprise. NSA wasn’t warned that there
would be a retaliation. We weren’t even able to readjust our [SIGINT] coverage in order to see the effects of the retaliation.”
⁷⁶

On August 7, 1964, Congress nearly unanimously approved what became known as the Gulf of Tonkin Resolution, which authorized
the president of the United States to “take all necessary measures to repel any armed attack against the forces of the United
States,” thus allowing the Johnson administration to expand the role of American military forces in Southeast Asia.

Postscript

This was an intelligence disaster of epic proportions. After all the available information is carefully reviewed and the arguments
on both sides given careful consideration, the overwhelming weight of the evidence now strongly indicates that there was no
naval engagement in the Gulf of Tonkin on the night of August 4, 1964.

Declassified documents reveal that President Johnson secretly doubted whether a naval battle had actually taken place. On
September 19, he kicked off a meeting of his national security advisers by telling them that he had “some doubt as to whether
there had in fact been any vessels of any kind in the area.” Despite his doubts, that afternoon the White House issued an
unequivocal statement that there had indeed been a naval battle that fateful night.
⁷⁷
As time went by, though, Johnson exhibited increasing doubt as to the veracity of the NSA radio intercepts that had been critically
important in justifying America’s entry into the Vietnam War. Years after the Gulf of Tonkin incidents, Johnson would occasionally
tease Secretary McNamara about the intercepts, chiding him with sarcastic jabs such as “Well, those fish [certainly] were
swimming,” or “Hell, those dumb stupid sailors were just shooting at flying fish.”
⁷⁸

This opinion is now shared by the two on-scene U.S. Navy commanders, Captains Herrick and Ogier (both retired), and even by
a repentant Robert Mc-Namara.
⁷⁹
Experts such as NSA deputy director Louis Tordella and INR’s Ray Cline have concluded that the intercepts were more likely
puffed-up North Vietnamese postmortem reports concerning the August 2 battle, rather than descriptions of the events that
allegedly took place on August 4
^.80

Even at NSA, there was much skepticism at the time about the veracity of the intelligence that the agency had provided that
justified America’s entering the Vietnam War. Frank Austin, the chief of NSA’s B Group, which was responsible for all communist
Asian targets, was, according to a declassified NSA history, “skeptical from the morning of 5 August,” as was Colonel John
Morrison, the head of NSA Pacific in Hawaii, who wrote a lengthy and critical analysis of the NSA reporting, questioning whether
an attack had taken place.
⁸¹
A declassified agency history of the affair notes, “The NSA analyst who looked at the traffic believed that the whole thing
was a mistake. The [intercepted] messages almost certainly referred to other activity— the 2 August attack and the Desoto
patrols. The White House had started a war on the basis of unconfirmed (and later-to-be-determined probably invalid) information.”
⁸²

It was not until 2000 that NSA historian Dr. Robert Hanyok wrote a detailed study of the Gulf of Tonkin incidents for an internal
NSA publication; it concludes, on the basis of a review of over one hundred NSA reports that somehow never found their way
to the White House, that the August 4, 1964, Gulf of Tonkin incident never happened. Hanyok’s conclusions are sobering: “Through
a compound of analytic errors and an unwillingness to consider contrary evidence, American SIGINT elements in the region and
at NSA [headquarters] reported Hanoi’s plans to attack the two ships of the Desoto patrol.

Further analytic errors and an obscuring of other information led to publication of more ‘evidence.’ In truth, Hanoi’s navy
was engaged in nothing that night but the salvage of two of the boats damaged on 2 August.” Hanyok’s controversial top-secret
report alleges that NSA officials withheld 90 percent of the SIGINT about the Gulf of Tonkin attacks in their possession,
and instead gave the White House only what it wanted to hear. He concludes that “only SIGINT that supported the claim that
the communists had attacked the two destroyers was given to administration officials.”
⁸³

But what ever doubts may have existed in August 1964 about the credibility of the evidence provided by NSA about the Gulf
of Tonkin naval engagement, in the end it really did not matter. It was no secret that, wanting to “look tough” in an election
year, Johnson administration officials were looking for a casus belli for attacking North Vietnam. So President Johnson, Secretary
of Defense McNamara, and the JCS appear to have cherry-picked the available intelligence, in this case SIGINT from NSA, in
order to justify a decision they had already made to launch air strikes against North Vietnam. Ray Cline stated that Johnson
and McNamara “were dying to get those air attacks off and did finally send them off with a pretty fuzzy understanding of what
had really happened.”
⁸⁴
The final word goes to an NSA historian, who concluded, “The administration had decided that expansion of American involvement
would be necessary. Had the 4 August incident not occurred, something else would have.”
⁸⁵

CHAPTER 7

The Wilderness of Pain

NSA and the Vietnam War: 1964–1969

A man’s judgment is no better than his

information.

—LYNDON JOHNSON, 1968

Flying Blind

Recently declassified documents make clear that everything we thought we knew about the role of NSA in the Vietnam War needs
to be reconsidered. One fact kept a secret until now was that after the North Vietnamese and Viet Cong converted all their
communications to unbreakable cipher systems in April 1962, as described in chapter 5, NSA was never again able to read any
high-level enemy communications traffic except for very brief periods of time. Throughout the war, the North Vietnamese and
Viet Cong constantly changed and improved their high-level diplomatic and military cipher systems, in the process killing
off the few cryptanalytic successes that NSA enjoyed. As a declassified NSA history notes, “it was not the sophistication
of Hanoi’s cryptography that hindered cryptanalysis, but the short shelf-life of its systems. Even then, the time between
intercept and decryption was still months.”
¹
At some point in the mid-1960s, NSA made the controversial decision to give up altogether on its efforts to crack the high-level
North Vietnamese ciphers and instead focus its resources on solving lower-level enemy military codes used on the battlefield
in South Vietnam and on traffic analysis.
²

Since NSA could not provide any high-level intelligence about the strategic intentions of Ho Chi Minh and the rest of the
North Vietnamese leadership, the U.S. government found itself repeatedly and unpleasantly surprised by the actions of the
North Vietnamese and Viet Cong. Failing to forecast the North Vietnamese–Viet Cong 1968 Tet Offensive was perhaps the worst
U.S. intelligence failure, one that occurred in part because, per a 1968 CIA postmortem report, “
high-level Communist communications” were “for the most part unreadable
” (italics added).
³

NSA’s best intelligence was derived from reading the diplomatic traffic of foreign countries like Brazil and Indonesia, which
maintained embassies in Hanoi. The cable traffic of foreign journalists visiting Hanoi was also a useful source of information.
For example, in 1968 NSA intercepted a message from a Japa nese journalist in Hanoi to his home office in Tokyo reporting
that he had interviewed and photographed a number of American POWs held by the North Vietnamese.
⁴

The North Vietnamese Enter the War in the South

Immediately after the Gulf of Tonkin incidents, the U.S. intelligence community tasked NSA with intensifying its SIGINT coverage
of both Viet Cong (VC) radio traffic inside South Vietnam and North Vietnamese Army (NVA) communications north of the DMZ.
The agency’s monitoring of VC Morse code communications traffic quickly identified a number of major enemy corps and division-size
headquarters staffs covering all of South Vietnam. NSA also began closely monitoring the radio traffic of the NVA unit that
ran the entire army logistics infrastructure in North Vietnam and Laos, the General Directorate of Rear Services (GDRS). GDRS
was a critically important target because it was responsible for moving men and supplies down the Ho Chi Minh Trail from North
Vietnam through southern Laos and Cambodia into South Vietnam.
⁵

Within weeks of initiating intercept coverage of GDRS, NSA began intercepting message traffic suggesting that elements of
a regular North Viet nam -ese Army unit, the 325th NVA Division, had begun preparing to cross into southern Laos from their
home base in Dong Hoi in North Vietnam. In November 1964, SIGINT confirmed that an enemy radio station operating along the
Ho Chi Minh Trail in southern Laos had suddenly converted its radio operating procedures to those used by regular NVA units.
A few weeks later, in December, CIA “road watch” teams in southern Laos spotted several battalions of regular North Vietnamese
troops moving down the Ho Chi Minh Trail in the direction of South Vietnam. In the ensuing months, traffic analysis coming
out of NSA tracked the movement of the 325th NVA Division through the Mu Gia Pass and southern Laos and into South Vietnam.
Although U.S. Army direction-finding assets confirmed the presence of this division in the Central Highlands of South Vietnam
in January 1965, the Military Assistance Command Vietnams’ (MACV) intelligence staff in Saigon refused to accept the presence
of NVA regular forces in the country because it had not been confirmed by POWs or captured documents. It was not until early
February that MACV finally agreed that the headquarters of the 325th NVA Division plus a subordinate regiment were in the
Central Highlands.
⁶

The Opening of the Ground War in South Vietnam

In South Vietnam, the ground war was moving into a new and more lethal phase. The initial landing of U.S. Marines took place
in March 1965, and by June the entire Third Marine Amphibious Force was operating in the northern part of South Vietnam, based
in the city of Da Nang. In July 1965, the first U.S. Army combat unit, the First Cavalry Division (Airmobile), arrived in
South Vietnam. As the number of U.S. combat troops in South Vietnam rose steadily, so did the number and intensity of North
Vietnamese and Viet Cong attacks. Forces on both sides began maneuvering for advantage, shadowboxing while waiting for the
other side to make the first decisive move.

The first battle of the new “American phase” in the Vietnam War began in August. Early that month, U.S. Army airborne radio
direction finding (ARDF) aircraft flying routine SIGINT collection missions over the northern portion of South Vietnam picked
up a heavy volume of Viet Cong Morse code radio messages coming from just south of the Marine Corps base at Chu Lai. By mid-August,
the ARDF aircraft had discovered the source of the Morse code transmissions and the identity of the Viet Cong unit sending
the messages. The transmitter belonged to the headquarters of the two-thousand-man First Viet Cong Regiment, which was secretly
concentrating its forces on the Van Tuong Peninsula, fifteen miles south of Chu Lai. The information was fed to General Lewis
Walt, commander of the Third Marine Amphibious Force, who immediately initiated a search-and-destroy operation against the
VC regiment. Designated Operation Starlight, it commenced on August 18. A marine battalion quickly penned the VC regiment
up against the sea, while another marine battalion landed on the peninsula and began wiping out the trapped Viet Cong forces.
By August 24, the marines reported that they had destroyed two battalions of the VC regiment, killing an estimated seven hundred
Viet Cong troops. On the negative side, over two hundred marines had been killed or wounded in the fierce fighting. Despite
the heavy casualty toll, NSA officials considered the success of Operation Starlight to be SIGINT’s most important accomplishment
in Vietnam up until that point.
⁷

Unfortunately, as was too often the case during the war, the use of body count metrics to measure success during Operation
Starlight produced a chimera. In fact, SIGINT showed that the majority of the First Viet Cong Regiment had somehow managed
to escape from the Van Tuong Peninsula. According to a declassified NSA history, radio intercepts showed that “within two
days of the battle, the First Regiment’s radio network was back on the air.”
⁸

Two months later, in October, three regiments of the 325th NVA Division launched an offensive in the Central Highlands with
the objective of cutting the country in half. In this first offensive in the south, NVA regulars scored a quick victory at
the Plei Mei Special Forces camp, twenty-five miles south of the city of Pleiku, but then were forced to retreat up the nearby
Ia Drang Valley when confronted by a strong force of American infantrymen belonging to the newly arrived First Cavalry Division
(Airmobile), commanded by Major General Harry Kinnard.

As the 325th NVA Division retreated deeper into the Ia Drang Valley, it was shadowed by five ARDF aircraft tracking the locations
of the radio signals of the division’s commander and his subordinate regimental commanders, which enabled Kinnard’s forces
to leapfrog up the valley in their Huey he licopters, harrying the retreating division every chance they got. At about four
thirty a.m. on November 14, a tactical SIGINT intercept team attached to the First Battalion, Seventh Cavalry, intercepted
a transmission indicating that a battalion of the 325th NVA Division (the Ninth Infantry Battalion of the Sixty-sixth NVA
Regiment) was trapped at the base of the Chu Pong Massif. Acting on this intelligence, at eleven a.m. helicopters dropped
the 450 men of the First Battalion, Seventh Cavalry, commanded by Lieutenant Colonel Harold Moore, at landing zone (LZ) X-Ray,
in front of the Chu Pong Massif, to destroy the enemy force.
⁹

But SIGINT can sometimes be wrong. As immortalized in the book and movie
We Were Soldiers Once . . . and Young
, Hal Moore discovered almost immediately that he was facing not an NVA battalion, but rather two full regiments of the 325th
NVA Division. Two days of fierce and bloody fighting ensued, much of it hand-to-hand. When it was over, both of the NVA regiments
had for all intents and purposes been destroyed, with the survivors retreating across the border into Cambodia. But the Battle
of LZ X-Ray, the first engagement of the Vietnam War between American and North Vietnamese troops, showed that the North Vietnamese
couldstand and fight against the better-armed Americans.

As in Operation Starlight, SIGINT’s per formance during the Battle of the Ia Drang Valley was not a complete success, with
a declassified NSA history reporting, “At least four times during the struggle, South Vietnamese and American units had been
ambushed by large communist units—twice during he licop ter landings—and SIGINT had been unable to detect the traps.” The
lesson learned from these two battles was that SIGINT was an imperfect intelligence source if used all by itself, without
supporting intelligence from agents, POWs, and captured documents. Sadly, as we shall see, this simple truth was forgotten
by later generations of senior U.S. field commanders in Vietnam.
¹⁰

SIGINT Successes in the Ground War in South Vietnam

While the Rolling Thunder bombing campaign in North Vietnam continued into 1966, in South Vietnam NSA was beginning to rack
up some impressive gains. The list of the agency’s targets grew rapidly in response to customers’ demands for more and better
intelligence, including information on the deployments and movements of North Vietnamese and Viet Cong forces down to the
tactical level, North Vietnamese fighter activities and surface-to-air missile locations and readiness levels, Soviet and
Chinese weapon and supply shipments to North Vietnam, North Vietnamese weather forecasts, civil aviation flights, and on and
on.

And despite its inability to crack the North Vietnamese military’s high-level ciphers, NSA was increasingly able to produce
vast quantities of intelligence about the North Vietnamese and to a lesser degree the Viet Cong forces operating inside South
Vietnam by cracking their low-level cipher systems, as well as making use of increasingly expert traffic analysis and direction-finding
data obtained by army and air force ARDF aircraft. Throughout the war, according to a declassified NSA history, “American
and Allied cryptologists would be able to exploit lower level communist cryptographic systems, that is, more precisely, ciphers
and codes used by operational and tactical-level units, usually regiment and below, on an almost routine basis. In fact, the
volume of the so-called low-to-medium-grade systems exploited by NSA was so great that by 1968 the exploitation had to be
automated.”
¹¹

This success quickly translated into better intelligence about the strength and capabilities of the enemy. A declassified
May 1966 Defense Intelligence Agency (DIA) order of battle estimate of the North Vietnamese military shows that SIGINT was
able to identify the locations of virtually every major North Vietnamese combat unit stationed in North and South Vietnam,
as well as the locations and complete aircraft inventory for every regiment in the North Vietnamese air force.
¹²

On the battlefield in South Vietnam, SIGINT quickly outstripped other intelligence sources in its ability to find and accurately
track the movements of the ever-elusive North Vietnamese and Viet Cong forces, which made destroying them immeasurably easier.
Jim Lairson, an army Morse intercept operator based at the huge Phu Bai listening post, in northern South Vietnam, recalled
an incident in February 1966, when the intercepts of the Viet Cong combat unit he was assigned to monitor began moving inexorably
toward his post. He remembered, “The [enemy] operator I was copying got frustrated with [his] control and switched from coded
to plain text. Our translator wasstanding behind me and as I typed Phu Bai on the paper. I got the word. There were three
battalions of Viet Cong coming at us.” The approaching enemy force was immediately hit by dozens of bombs dropped by an on-call
force of U.S. Air Force fighter-bombers, and the threat to the base passed.
¹³

One of the most skilled users of SIGINT in Vietnam was Major General William DePuy. Commander of the First Infantry Division,
based north of Saigon, he owed his skills largely to his experience in the intelligence field before coming to Vietnam. In
July 1966, army ARDF aircraft located the headquarters of the 272nd Regiment of the Ninth Viet Cong Division near the village
of Minh Thanh, in Tay Ninh Province near the Cambodian border. In the resulting battle, troops belonging to DePuy’s division
surprised the Viet Cong regiment, killing three hundred VC soldiers and putting the entire Ninth VC Division out of action
for the next three and a half months.
¹⁴

Three weeks later, in August, U.S. Air Force EC-47 ARDF aircraft flying over Quang Tri Province, in the northernmost part
of South Vietnam, intercepted the largest number of NVA transmitter fixes found in the DMZ since America’s entry into the
war. The radio emitters belonged to the North Vietnamese 324B Division, which was in the process of trying to flee back across
the DMZ into North Vietnam after being mauled by U.S. Marine Corps units earlier that month. B-52 bombers were called in to
plaster the locations of the 324B Division with carpet bombing. Hundreds of NVA troops died in the resulting conflagration
of high-explosive ordnance and napalm. The director of intelligence of U.S. Pacific Command reported on September 29, “Without
[EC-47’s] work and that of more sensitive intelligence [SIGINT], we would be completely in the dark about the enemy situation
in the DMZ.”
¹⁵