War of Numbers (32 page)

In a few days’ reading, the extraordinary story of what the communists had been up to in Cambodia began to emerge in detail. In the months after Sihanouk’s fall, the Vietcong invaders had assembled a large and complex advisory system, bent on fielding a Cambodian rebel army to compete with Phnom Penh’s. Some documents told of chains of command, of radio nets, of hospitals and training schools. Others mentioned artillery companies, commando battalions, infantry regiments, even a division. The dimensions of the new army were unclear, but the number of documents and the complexity of the system they described suggested it was very big. I assumed its existence was well known.

I found shortly that it wasn’t. In mid-May, a just-published CIA memorandum about Cambodia arrived on my desk; it stated that the number of Khmers in the communist military structure in Cambodia lay in the range of five thousand to ten thousand. In light of what the documents said, the numbers seemed so uncommonly small that it struck me at once that something was wrong. Where had the numbers come from? I set out to inquire.

The DDI had three offices that dealt with the war in Cambodia. Neither the first nor the second could say where the numbers arose; furthermore, they asserted, they didn’t get the documents. The third office was the Manpower Branch, which still had the job of tallying communists in Cambodia. The branch got the documents all right, but apparently hadn’t sorted them out. Asked for its files on the Khmer communist army, a branch analyst proffered a thin folder that contained a dozen or so low-level agent reports—but no documents. There was also an internal memorandum, dated late April 1971, which stated the branch had done virtually no research on the Communists in Cambodia for almost a year. A check of CIA cables from Phnom Penh, Saigon, and Bangkok turned up the same story: no research on Cambodian numbers.

Thereupon I called the Defense Intelligence Agency in the Pentagon and asked for its Cambodia shop. A Sergeant Reisman came to the
phone and said he was the person charged with keeping files on the Cambodian rebels. The range of five thousand to ten thousand, he declared, had come from the Cambodian government’s army intelligence (G-2) a year earlier. DIA had tinkered with the number, then left it alone because it seemed “reasonable.” DIA had never done an in-depth study on the matter, he said. “There’s no one at DIA to do a study,” he declared; “they hit us with too much shit.” I thanked the sergeant and hung up. The time was five o’clock in the afternoon of 28 May 1971, a Friday.

So that was the story. U.S. intelligence had neglected to inquire whether our enemies in Cambodia had raised an army. I had happened upon it while looking for something else. Four hundred and thirty-six days had passed since 18 March 1970, the day Cambodia joined the war.

I began in earnest on Saturday morning to write a paper gauging the size of the Khmer communist army. Since the evidence was at hand—hundreds of captured documents, prisoner-of-war interrogations, and CIA field reports—the next step was to determine whom to count. I merely followed the criteria of the Cambodian G-2, whose range of five thousand to ten thousand the year before had included the communists’ three types of soldiers: those in the main forces, the local forces, and the guerrilla-militia. They are the same categories the communists have in Vietnam and Laos, and, according to the French army study exhumed from the archives, the same ones the Khmer rebels had in 1953.

Writing the first draft took a week. It dealt with the categories one by one, and, using the same methods the CIA had earlier employed in estimating Vietcong numbers, came up with separate figures for each type. On Friday, 4 June 1971, the three estimates were complete, the line drawn at the bottom, and the total arrived at. They were:

This sum, the draft observed, was from ten to thirty times higher than the one carried on the U.S. government’s official books. The reason for the discrepancy, the draft pointed out, was that American intelligence had missed the phenomenal growth of the rebel army from not having looked.

The draft explained at length who was in each category. The main forces, it said, were the communists’ mobile reserve, and consisted of large numbers of independent companies and battalions and about a dozen infantry regiments—all but one of which the CIA memo of May had missed. The local forces, akin to the U.S. National Guard, fought in each of Cambodia’s nineteen provinces and more than one hundred districts. The guerrilla-militia were local partisans who defended communist-held villages and hamlets. Poorly gunned in most places, they were armed to the teeth in others. The Vietcong guerrilla-militia had caused about one in four of all U.S. casualties in South Vietnam.

Experience told me that the Cambodian draft might run into heavy weather. As the CIA’s only analyst on Vietcong strength in 1966 and 1967, I had written a series of studies presenting evidence that the Vietnamese communist force in South Vietnam was closer to 600,000 than the 280,000 then carried officially on U.S. books. The studies fared badly. Some were pigeonholed, others killed outright. But the evidence they advanced was so abundant that it was hard to overlook, and pressure mounted throughout 1967 to raise the official estimate.

American intelligence chose instead to rearrange the books to keep the Vietcong numbers low. Following guidelines form General Westmoreland’s headquarters in late 1967 to maintain the Vietcong estimate below 300,000, his intelligence staff used two methods to do so. The first was to remove from the books categories of Vietcong who had been there all along. The second was to employ “conservative” techniques—such as omitting troops in small units—in accounting for types of soldiers supposed to be on the lists. And instead of going up, the Vietcong estimate went down—to about 240,000.

When the Tet Offensive hit in January 1968, large numbers of soldiers missing from the estimates popped up in the middle of South
Vietnam’s cities. Two months later, the CIA, which had winked at the military’s pre-Tet wiles, reversed itself and vowed officially that there might be as many as 600,000 Vietcong. But it took Tet to get the agency to reassess, then accept the evidence of 1966, an exercise, an analyst later wrote, “in counting Indians after they’d come through the window—not so much intelligence as a sort of morbid curiosity.”

Fearful that my Cambodian draft might go the way of some of the papers before Tet, I determined to see that it got widely read within the government, even if it meant handing out copies myself. Before leaving the CIA that Friday, I told Bud, deputy chief of the staff to which I belonged, that I had written a paper on the Khmer communist army, and would do a final draft over the weekend. He looked up distractedly from his typewriter and nodded.

The final draft, handwritten, was complete one Sunday evening in June 1971. It took most of that Monday to type the thing up. When done, it came to about forty pages, with a hundred or so footnotes that referred to some twice as many sources. Late that afternoon I called an acquaintance on Kissinger’s staff and told him of the paper’s existence. He asked for a bootleg copy. That evening I stopped by the Executive Office Building (next to the White House) and handed him a Xerox copy in a manila envelope.

Arriving at work on 8 June, I gave the original to my boss, Mr. Pontiac, the staff chief, to whom Bud was deputy. Mr. Pontiac said that he would read it later. I went upstairs and showed the draft to various functionaries who dealt with Cambodia, including George Carver, chief advisor on Indo-China to the CIA’s director, Richard Helms. Mr. Carver glanced it over, declared that 100,000 to 150,000 seemed to him too high, and suggested that “45,000 to 50,000” sounded better. When I returned to my office, Mr. Pontiac, who by this time had read the draft, said that under no circumstances was it to leave staff channels.

Mr. Pontiac spent a good part of the next day discussing the paper, I was told, with the DDI hierarchy, including its deputy, Mr. Walsh, and perhaps its chief, Mr. Proctor. After the consultations Mr. Pontiac called
Bud and me into his office. It was 3:00
P.M.
He declared the paper was “unconvincing,” that it was written “hurriedly and in anger,” that it was “lacking in integrity,” and that, moreover, my writing it had been a “clandestine operation.” Were I to write another like it, he asserted, I would find myself “out on the streets.” When he saw me taking notes on his remarks, he told me to stop. Whereupon he instructed Bud to come up with a more acceptable draft, and dismissed us both. I said to Bud as we left that I didn’t think the draft was a clandestine operation because I’d told him about it the Friday before. “Yes,” Bud replied, “but you didn’t tell about the troll under the bridge.”

The Kissinger staffer called on the tenth to say that he and a colleague had read the draft and thought it a good job. He suggested sending a bootleg copy to the Defense Department, which was apparently unaware of the problem. The following day Mr. Pontiac convened a meeting at 11:00
A.M.
to discuss the draft. I was not invited. Nine other CIA officials who worked on Cambodia showed up. Mr. Pontiac told them that the draft was for “internal use only”—it was not to leave the building—that they couldn’t see the original but would get copies of Bud’s revised version when it came out. He also stated that the DDI front office had decided to turn over the revised version to the Manpower Branch for “action” since the branch was still the agency’s official tabulator of communist numbers.

When Bud completed the revised version, it had two main changes from the original: the phrase “100,000 to 150,000” was amended to read “as many as 100,000, or more,” thus making 100,000 seem the midpoint rather than bottom end of the estimate; and all statements which implied that the CIA hadn’t done its homework were deleted. Otherwise unscathed, the revised paper made its way, footnotes and all, to the Manpower Branch.

On 18 June, the Manpower Branch chief, Mr. Tate, assigned a newcomer to the branch, Herman Dowdy, to supervise the Cambodian project. Unschooled in the labyrinths of communist army organization, Dowdy had never before worked on estimates of enemy strength. A few
days later I received an under-the-table copy of Dowdy’s first written foray into the business of estimating communist numbers. It was a memorandum commenting on Bud’s revised version. The memo stated that the revised version was structurally unsound, and that it wrongly implied that the Khmer rebel army offset the government’s.

Early in the morning of 22 June, Mr. Dowdy told a Manpower Branch analyst that neither earlier draft “would ever see the light of day.” This was officially confirmed at a meeting, chaired by Mr. Tate, at 10:30
A.M.
on the CIA’s seventh floor. Mr. Tate announced his branch was going to scrap both early drafts and write an entirely new one. Asked how long it might take, Tate would not say.

Not long after the 10:30 meeting, I plodded disconsolately down a gray stairwell to the Manpower Branch to see what could be salvaged. As I arrived, a branch analyst named Peter Snider emerged from Tate’s inner office. Tate, he said, had just put him on the Khmer communist army project under Dowdy’s supervision. Like Dowdy, Snider was new to the branch, and had had little experience in dealing with communist numbers.

I asked Snider if I could help. He replied that he needed the references—documents and so forth—mentioned in the footnotes of the just-killed paper, since the Manpower Branch hadn’t yet got them. I offered to show him where they were in the archives.

Snider then sat down at his desk, and remarked casually that his forthcoming study was unlikely to come up with more than thirty thousand Cambodian Communist soldiers; nor would it find less than ten thousand, he said. He stated that one of the reason for this was that his study would include only two of the Communists’ three types of soldiers: the main forces and the local forces, but not the guerrilla-militia.

Perplexed, I began to advance reasons for keeping guerrillas in the estimate. In the first place, I said, the Cambodian G-2 had included guerrillas in its year-old range of five thousand to ten thousand, and for the CIA to drop them now would mean that Washington and Phnom Penh had separate definitions of who the enemy was. Then, of course,
the Manpower Branch had included VC guerrillas in its Vietnam estimate, and the Cambodian guerrillas performed exactly the same tasks next door. And finally, they were a clear military threat—witness our losses to Vietcong guerrillas—since thousands bore arms, and nearly all were trained how to fight. But Snider shook his head and said they were too “amorphous” to count.

He then stated that he wasn’t going to include service troops either. Again taken aback, I declared that ordnance sergeants, medics, and such like, where were an integral part of the main and local forces, were just as valuable as combatants: in some cases more so, since they took longer to train. Besides, I said, when the CIA reported the size of Phnom Penh’s army, it always included the government’s service soldiers—so why not the rebels? Counting them on one side and not on the other distorted the odds, since almost half of Phnom Penh’s force consisted of service troops. Snider repeated his resolve to exclude rebel logisticians.

By this time I was filled with misgivings. Here was another case, I suspected, of the game before Tet: supply subordinates with a maximum number of enemy soldiers to count and let them figure out how to do it. In late 1967, it was three hundred thousand Vietcong. Now it looked like thirty thousand Cambodian rebels.

So I asked Snider if Tate had given him the number. Snider shrugged but did not answer. (But when a friend of mine who worked in the Manpower Branch later asked Snider the same question, he answered simply, “Yes.” I repeated the query 10 May 1973, just before I quit the agency. Snider told me that Tate had given him the components to count and that in his opinion this amounted to the same thing. He also said that his immediate boss, Dowdy, had told him that morning that his job was to “refute” the numbers in the earlier drafts.)