Cloud Nine: A Problem in Intelligence Production
APPROVED FOR RELEASE 1994
CIA HISTORICAL REVIEW PROGRAM
2 JULY 96
Current intelligence delivers a maximum effort.
CLOUD NINE: A PROBLEM IN INTELLIGENCE PRODUCTION
James W. Featherstone
On the 24th of January, 1969, there descended upon the Directorate for Intelligence at CIA a document that for a brief but frantic period was to try the resources of the Directorate in some ways more severely than they had ever before been tested. This document was a National Security Council directive, later dubbed by those who became affected by it, with appropriate irony, "Cloud Nine." This directive called for an inventory of the international situation as of 20 January 1969, in the form of a "current assessment of the political, economic, and security situation and of the major problems relevant to US security interests and US bilateral and multilateral relations" world-wide. It demanded, in addition, "a discussion, where appropriate, of the data upon which judgments are based, uncertainties regarding the data, and alternative possible interpretations of the data." To make certain that the response was properly pointed, the directive posed, in 52 pages, a total of 893 probing questions touching almost every country on the globe. The answers were to be in the President's bands by the 20th of February, a matter of 26 days, including weekends.
Obviously this was a task of formidable magnitude, one that at first glance appeared almost impossible of fulfillment in the time span allotted. Of course, certain short cuts could be taken. It is always tempting, for example, when confronted with a requirement of large dimensions, to look on the shelf for already canned material that could be dusted off, updated if necessary, and ladled out lukewarm to the consumer. Another timesaving device is to by-pass the usual processes of editing and review, sending forward immediately from the typewriter the analyst's sweaty draft. Unfortunately, although in many instances, at least, the analyst views the draft as perfect, the editor thinks it is merely perfectible, and his ministrations toward that laudable end take time.
In this case, it was decided that there would be no short cuts. A really fresh look at the situation would be taken, candid judgments would be rendered, and the answers would be made as incisive and unequivocal as it lay in our power to make them. Each answer would be responsibly edited, and the entire lot would be reviewed by two senior officers—one of them the chairman of the task force established to complete the project—who would ensure that, stylistically and substantively, the whole massive response met the high standards set by the DDI. In appearance the response was to be neat and attractive, if not as psychedelically colorful as a Madison Avenue product, and it was to be carefully packaged. And finally, in the words of the Director, as far as CIA was concerned, at least, the deadline would be met. His determination in this respect may have been influenced somewhat by knowledge that the directive had been levied also on the Departments of State, Defense, and Treasury, putting the Agency, in a sense, in competition with them for timely production of a quality product.
The manner in which inter-Office and inter-Directorate teams were manned and organized to prepare the answers and get them edited and coordinated is a complex story in itself. Here, however, we propose to deal with a different aspect of the Cloud Nine problem, one with which we are well acquainted especially in current intelligence production, but which is not generally thought of until it becomes potentially a major obstacle, as it did in the case of Cloud Nine. This aspect embraces the mechanics of reproduction, the process of getting a clean draft typed, a final draft printed, and a finished product ready for distribution to meet a deadline.
Anyone with any experience in the management of current intelligence is likely to be more familiar than be likes with the sentence, "It's in the typewriter." This sentence, usually uttered in a tone of bored indifference by a writer who has been asked what has happened to an item that has a rapidly approaching deadline, typifies an all too common feeling among professionals that once the draft is written, the work is done and the project completed. Would that it were so; our task would be much easier. Few, of course, would buy the proposition that the clerical process is as difficult or complex or even as valuable as the analytical/writing process. But the fact remains that in many instances, unless the analyst's work is typed and reproduced and distributed at the right time, the analyst might just as well not have done it at all, for all the effect it will have. So from this point of view, the clerical process, whatever else its level in the scale of values in the profession of intelligence, is just as essential as the analytical one. And it is played down or disregarded only at the peril of derailing the entire train of intelligence production.
In the case of Cloud Nine, it was clear from the outset that the production process, unless carefully, managed, would unquestionably defeat us. This was partly the result of the sheer bulk of the project. It stemmed also from the decision to start from scratch and to apply the full editorial review machinery to the final product. It was closely related, obviously, to the shortness of the deadline: less than three weeks remained after the writing teams had been organized in which to finish and deliver the product. And, of course, the fact that all of the production personnel involved in Cloud Nine had other duties that normally occupied them full time, duties that still had to be carried on simultaneously, compounded an already difficult situation. Given all of these factors, it was legitimate to question whether the deadline would, in fact, be met as we plunged into work
It was estimated originally that CIA's response to the NSC directive would require about 1,000 pages, and in actuality this estimate was only slightly below the mark. Each original contribution would be subjected to at least two levels of editing. No one, least of all the analyst, was hopeful that minimum alterations in the first or any subsequent drafts would suffice. The average editor is said to live by the principle that the pen is mightier than the sword, and it is commonly held that he forgets that both can be equally lethal against their natural targets. So it was conceivable that every contribution to the aggregate response would have to be typed a minimum of three times and quite possibly, when finally reviewed, for a fourth. The massive size of this typing load made it potentially a serious obstacle to our meeting the deadline, a built-in impediment at every stage of production between the writer, the reviewer, and finally the print shop. A backlog of any consequence at any stage of the process could cause us to miss the target date by a wide margin.
In fact, the key to the success of any large operation of this sort is to establish and adhere to a realistically phased writing-typing correction schedule that will ensure a smooth, steady, and controlled flow of material from the writers through the editors to the printers. This began, in the case of Cloud Nine, with scheduling the actual writing so that the answers to the easier questions would be prepared and edited first and the more difficult ones deferred until close to the deadline. It was hoped, and indeed it turned out to be the case, that the senior reviewers would be confronted on any given day with no less and if possible very little more than they could plow through in around ten hours. Since these two gentlemen constituted a marked constriction in the pipeline, having to read everything, it was mandatory that their part of the operation proceed smoothly.
To ensure the essential control of the flow of material, a control center was established in OCI's Publications Support group. Here a number was assigned to each question in the NSC directive (they were unnumbered in the document itself) and notation made of the team and office responsible for producing the answer and of the scheduled date of delivery to the senior reviewers. As the various answers were written, edited, and typed in second draft, they were sent to the control center. The center, in turn, passed them to the senior reviewers. When the latter had finished massaging them, they were returned to the control center, which then had them put on mats in final form and sent them to the print shop for reproduction. The control center could tell at any time, and kept the Task Force Chief informed, whether the production schedule was being maintained, where any particular answer was at any time, etc. The system made it possible also to have the individual answers printed as they were completed, regardless of sequence, and to ensure that assembly and pagination of the entire response could be done in orderly fashion at the very end of the process. Obviously it would have been fatal to hold the drafts until all parts of the response were in and had been put in the sequence required by the NSC directive before beginning to print them.
The production schedule was extremely tight, and there was little latitude for slippage. We were fighting the clock to such an extent that, rather than hand carry, we used the pneumatic tube system to get papers from one part of the building to another without delay, and to keep everyone busy. At one point, the tube system broke down, causing a certain amount of panic until the missing papers were located and extracted, and business could go ahead again.
Despite this and a few other lapses, our attack on the Cloud Nine problem was well-organized, and it proved to be effective. The flow of manuscripts actually began earlier than expected, and no large backlog developed at any time. But even with the best organized and most rational schedule, we would still have failed to deliver on time if we had not had the MTST—Magnetic Tape Selectric Typewriter. We had eleven, and sometimes twelve, of these machines at our disposal, five of them in OCI's Publication Support group and the remainder elsewhere in the Intelligence Directorate. We could have used more, since not all of the substantive teams had one available to them. The answers to the vast majority of the questions, however, were put on magnetic tape after initial editing and review had been completed by the substantive teams. The tapes and the corresponding typed drafts were then delivered to the Control Center in OCI, where the tapes were retained while the senior reviewers were working independently on their copies of the drafts. The revisions made at this level were then, with the aid of the MTST, incorporated in a finished, corrected tape and a final draft typed simultaneously.
The advantage that the MTST has over the ordinary manual or electric typewriter, and its unique contribution to Cloud Nine, may not be apparent from this generalized description of procedures. So to be a little more specific: At a reasonable estimate, probably less than five percent of all the pages in the analysts' original typescripts were completely untouched in both levels of review. Retyping all of these pages by hand would have been enormously time-consuming, not to mention the strain it would have imposed on the typists themselves. But when a draft on tape is run through the machine, the MTST retypes automatically at a very high rate of speed. The trained operator stops the machine wherever a correction is necessary, enters the change on the new typed draft and the corrected tape, simultaneously, and directs the machine to proceed until the next change is encountered. By this means a clean, corrected draft, together with a new tape, is rapidly produced with a minimum of effort compared with what would be required without the MTST. Depending upon which stage of production we are talking about, this new draft is either a typescript designed for further review or a mat for printing, if final review has been completed. We believe that without the MTST, it would have taken about twice the number of typists twice the time to do the same job.
A final factor bearing on our success with Cloud Nine—intangible but nonetheless critical—was the morale of the machine operators.
Theirs was an exacting task, and, despite the magic of the MTST, a nerve-racking one because of the constant pressure on them to get the thing done, and done accurately. Their performance was splendid, their pride in their work obvious and justified. At one point when the Task Force Chief inadvertently referred to them as typists, he was politely but firmly told that they were machine operators. The distinction is a valid one. The girls do have to be trained to operate the MTST, and certain stages of their work are somewhat akin to computer programming.
Between February 3, when Cloud Nine drafts began to pile into Publication Support's office, and February 16, when the mats were finished, the machine operators worked every day. The five in the Publication Support group, alone, together with three proofreaders, racked up a total of 378 hours of overtime. Their supervisor had to watch them closely—a pleasurable occupation in any event—for signs of exhaustion and falling efficiency. Occasionally he had to tell one to pack up and go home. But he was most impressed by their energy and drive and their devotion to duty, which clearly went beyond any desire for overtime compensation.
Once the mats were done, the only remaining hurdles to be surmounted were printing and assembling the answers. These tasks were performed in Printing Services Division's plant on the seventh floor of Headquarters Building, operating around the clock Considering that the various parts of the project were delivered to the plant piece-meal and were printed on arrival, without regard for final order of the answers, the job of collation alone was staggering. Small changes in the texts, moreover, were being introduced up to the very last moment. Printing Services Division designated a Control Officer for Cloud Nine who, in close coordination with the Control Center in OCI, monitored the flow of mats into the seventh floor plant and ensured that the printing schedule was maintained. High standards of appearance were demanded and met. And, of course, in this as in other Agency components engaged in Cloud Nine, regular routine work requirements had to be carried on at the same time.
When Cloud Nine was finished and put together, it filled 7 volumes and 1030 pages, counting inserts. The product was the result of a team effort probably unparalleled in the Agency's history, involving cooperation—smooth at all times—between elements of the DDP, ONE, DDI, DDS&T, and DDS. The Director called it the most extensive compilation of intelligence data and judgments ever assembled in such a brief time span. It was delivered to the White House a day before the deadline, and could actually have been sent over two days early. But that would have been ostentatious and might, besides, have promoted the dangerous view that Agency deadlines can be clipped back more severely than they already are.