151. "A Management Computer for the President," Technology Review, Vol. 86, No. 1 (January 1983), pp. 38-45.


In July 1979, a New York Times reporter called the White House press office with what seemed at first a routine query. A year earlier, he said, the president had promised to study the role of coal in solving our energy crisis and report to the nation within 30 days. (President Carter was fond of setting such deadlines for himself and his staff.) The reporter wanted to know what had happened to the promised study. The press officer did not know. He called around to other White House agencies, to no avail. He checked with the Department of Energy, but no one could recall the president's charging the department with such a task. Eventually, much too late to help the embarrassed press officer, it was established that the Interior Department was working--still working--on the study.

This incident was only one of many that helped foster a sense of lack of presidential follow-through. (Indeed, this is a problem by no means limited to one administration.) It also underscored the fact that the president had no routine procedure to keep track of what he assigned people and agencies to do, no expected completion dates, no list to show who is accountable, no systematic follow-up to establish who is behind schedule. In short, the president did not have a comprehensive management tracking system at his fingertips.

To remedy this situation, an attempt was made to use a computer to provide such a tracking system. I participated as an adviser--and I saw the efforts derailed by serious technical and political problems. These problems deserve attention not only because they limit the president's effectiveness but because they hinder the use of computers by managers everywhere. Indeed, at one point I asked computer expert Richard Garwin of IBM if the top executives of his internationally renowned computer-producing firm used computers for management tracking. His answer was very brief: no.

One major difficulty in providing the president with a proper tracking system was the issue of what constitutes a presidential directive. For instance, if during a speech the president reassures a group of American exporters that he has every intention of helping them increase exports next year, and that he will seek to open 20 new trade offices overseas for that purpose, does this constitute a directive? To whom? Should a management tracking system encompass only those matters the president orders personally--or also the many other instructions issued in his name by the presidentially appointed White House staff?

Top business executives face a similar problem, albeit on a different scale. They, too, deal with the outside world. For example, announcing that the corporation will do what it can, with all deliberate speed, to hire and promote more minorities and women may "buy" the corporation or its head some time and political points. Not to put those intentions into the tracking system may make the statement into an open lie, signaling that no implementation is expected. Putting them into the tracking system literally may trigger much more action than desired. Politics thrives on ambiguities; computers on ferreting them out and rejecting them.

There was also a highly charged authority and "access" issue: who would be in charge of such a White House tracking system? Note that this person would in effect oversee the performance of all the president's men, women, and agencies. Even what seemed like a minor question--where to place the computer-- turned out to be tricky. In the White House? It would smack of 1984. Outside? Where? (In corporations this issue will take the form of asking if computers should be a staff service or a headquarters function, and will "the" computer be open to direct access by managers of local plants and subsidiaries?)

A Many Splintered Thing

Most people who refer to "the White House" think about it as composed of the president and his staff. Actually it is organized into divisions, each with its own head, surprisingly strong divisional loyalties, and considerable infighting. The names and duties of the divisions change from administration to administration but not the basic organization. In the Carter White House, a little noted division was the Office of Administration, which took care of running the place internally, from writing checks to printing the president's speeches. It also managed a library and a research staff responding to questions from other divisions. The head of the Office of Administration, who also served as special assistant to the president for information (information flowing into the White House, not the other way), was a Georgian named Richard M. Harden. He retained me as "senior adviser," occasionally drew upon my counsel, and frequently offered it to other divisions. Developing a management tracking system was one more service the Office of Administration was going to render to the White House.

From the start, another matter familiar to politics deeply affected the way that the tracking system did not get underway. By mid-1979, the election loomed on the horizon and there was a growing demand for a list of the achievements of the administration. This was a more delicate matter than it might first seem. By law, public funds and resources cannot be used for election purposes, and this prohibition encompasses the White House staff. Indeed, people who openly worked on the reelection campaign had to be "Hatched"--removed from the public payroll under the provisions of the Hatch Act--and moved to Democratic Party headquarters. Nevertheless, the Carter administration, like others, tried within the limits of the law to stretch matters a bit, to use the White House staff for activities that might be in the public service as well as useful for the campaign. The assignment of compiling what the president had done was a case in point.

Twelve young persons, some regular White House staffers and some interns--all "politically reliable"--had been brought together under the stewardship of Earl Bender. They worked in the Old Executive Office Building, located next to the White House, and on my first day of work I visited the group. Having had some 20 years of experience in collecting and processing data, I was taken aback by the complete inexperience of the crew and its leader. Bender was an excellent campaign worker, with a keen sense for partisan politics and precinct organization, but he had little knowledge of the problems of data processing. The rest of the staff was less qualified. What the staff was doing was listing next to each other highly divergent items. These included statements culled from presidential speeches, legislation sent to Congress but not necessarily enacted, actual accomplishments, some steps undertaken by the White House staff under the president's directives, and some steps by various federal agencies on the initiative of their heads.

Even a brief review of the lists the group churned out showed clearly that the potpourri they were blending neglected any consideration of two matters that all compilers of information, political or otherwise, must face--the level of reality and the level of abstraction. The level of reality concerns the difficult transition from expression of goodwill and intention to actual performance, a problem familiar to anyone who ever decided to stop smoking or lose weight. It is of course much more complicated in the White House, since promises, gestures of support, and threats play as large a role in politics as actual performances. Also, to list only accomplishments, Bender said rather wryly, would not make a very long or compelling list. So "everything" was included, except some steps taken by agencies--especially the Department of Health, Education, and Welfare under Secretary Joseph Califano--that were clearly incompatible with White House policy.

Information Without Knowledge

The second matter the group overlooked, the level of abstraction, is an even greater issue. I had faced it through much of my career as I tried to develop a social-science theory of social change. When attempts are made to engineer social change, one of the factors that determines success is the amount and quality of the knowledge available about the nature of the problem and possible correctives. But before one can determine the amount and quality of the knowledge involved, one must decide what constitutes knowledge as distinct from information.

Most people tend to think of knowledge as a compilation of facts, somewhat like a telephone directory, with one bit of information listed next to another, ready to be used. However, such bits of data are often not usable by themselves. For one thing, their meaning--unlike information in the phone book--cannot be established without understanding the context in which these bits exist. The context contains the assumptions implicit in the facts and is the source of their meaning. To provide a simplistic example, the information that the temperature is 30 degrees tells us nothing unless it is specified whether it is Fahrenheit or Celsius, which defines the scale on which the point 30 finds its meaning. Human knowledge in general and science in particular are crisscrossed with such contextual assumptions, only the contexts in which the bits find their definitions and their meaning are much more complicated, richer, and more consequential. Thus, for example, the same national statistics will lead a Keynesian economist to rather different conclusions from those who subscribe to supply-side economics. Keynesian or supply-side theory provides the context.

Another limit on the value of bits of data is that no single observation is usable in isolation. Knowledge is gained by collecting a variety of observations, often partially incompatible, and synthesizing them into conclusions based in part on the observations themselves, in part on past observations, and in part on the logic of argumentation. This process of knowledge-building often leads one to play down some observations and play up others. Without it, isolated facts are misleading rather than informative.

I cannot stress enough that, as I see it, not only does science work like this, but also all kinds of knowing. For instance, in evaluating a presidential candidate or deciding whether to join or leave a political party, citizens cannot and do not relate to thousands of facts--that they like the candidate's policy on El Salvador but not his national security adviser, they favor his position on American fishing rights but not on the law of the seas, and so on. They tend instead to form a limited number of contexts, in part on the basis of general predispositions and in part on the basis of specific observations. For example, they may form an economic context, one on foreign affairs, and a general impression of personal competence.

Regarding the economy, for instance, citizens may feel that their first concern is to fight inflation, second to compete overseas, third to keep jobs. In this context, the Consumer Price Index (however misleading) becomes of paramount importance, unemployment data less so. Many other facts, such as decline in labor mobility, are either meaningless "bits without a home," or gain meaning once they are related to one of the core of contextual assumptions. (For example, reduced labor mobility may be seen as bad for economic strength and hence international competition.)

Institutionalized Incapacity

Armed with this understanding of how knowledge is formed, I was convinced that the telephone-directory approach worked out by Bender and company would be of no avail. They were listing items, some sheer facts (for example, the president appointed x women judges and y black judges), and some contexts (the president has been firm in his support of Israel) as if they were all of the same level of abstraction. However, Bender said he was quite willing to try his hand at integrating the information into meaningful packages. So I began working on this project, in parallel with efforts to develop the generalized computer tracking system.

We figured that the best approach was to establish a small list of contexts, write for each what we came to call a "preamble" summarizing the contextual assumptions, followed by key facts rather than a total list. Samples of our efforts were reviewed by Harden and, informally, by staff members of other White House divisions. Their reactions varied from receptive to enthusiastic. Soon, several divisions suggested additional facets they would like to see covered. "Congressional Liaison" felt it would be useful to have geographic information, so they could tell a congressional representative--at the push of a button--what the president had done recently for his or her district. A young staffer working for Anne Wexler, chief political adviser to the president, suggested listing also the main interest groups that support or oppose measures being promoted by the White House, to help in lobbying efforts.

Things seemed to be going well, but then we ran into institutionalized incapacity. This problem was identified years ago by sociologist Thorstein Veblen, who pointed out that a specialist often does not see what any straight-thinking person can; the specialized mind, honed in one direction, acquires an inability to perceive matters that are strung in a different direction. Above all, specialists of different disciplines find it difficult to communicate with each other because of the implicit contextual differences among disciplines. Thus, for instance, when scientists testified before a congressional committee that they thought there was a very high probability that a test ban on nuclear weapons could be effectively enforced without on-site inspections (to which the Soviet Union vehemently objected), they felt that they had made a strong statement in favor of the ban. Scientists tend to think in probabilities and recognize practically no absolutes; "very high probability" is about as certain as things ever get. In contrast, many politicians are by training lawyers: if they find loopholes, this spells trouble. Thus, the scientists' statement about the test-ban treaty suggested to members of the congressional committee that it had enough loopholes to make it unreliable and sent them looking for ways to close every one before ratification. The issue here is not who was "right" but the difference in perspectives.

Well, Bender was reporting to a lawyer who was the deputy of another lawyer, and while the head man was quite able to see the forest for the trees, the deputy was much more committed to the fine print than to contexts. He strongly and repeatedly vetoed any notions of preambles or contexts and instructed Bender to stick with "the facts." This the group did, and the usefulness of its efforts turned out at best questionable.

Knowledge for Management

Despite this setback, work on the management tracking system continued. Indeed, it occurred to me that the great effort that went into collecting information about past performance could serve as a basis for following evolving "future" events. So I suggested to Harden that we should put into a computer all actions that had been carried out by the White House and the agencies, organized in contexts according to subject areas (such as energy, health, education) and subareas (solar, nuclear, coal). This would include a list of the people accountable and, in the case of ongoing projects, the expected completion dates. We could then update the system daily as new presidential directives were issued, thus providing a detailed progress report on White House activities. But it didn't quite work out quite that easily.

The first thing Harden and I found was that the White House did not have an appropriate computer and feared having one. This statement requires some elucidation. While the White House had several computers, these were deliberately either small or specialized. The Secret Service had one, but strictly for its own purposes (such as keeping track of all visitors). Hugh A. Carter, special assistant to the president for administration, also known as "Cousin Hugh," had one, but it served mainly for housekeeping purposes. The Office of Management and Budget and the National Security Council each had their own. But none of these was suitable to, or available for, general management tracking purposes. Moreover, the very idea of creating a central computerized system to shadow the work of all the government agencies smacked too much of Big Brother. "What will the press say?" was a rhetorical question often raised. And the fact that feeding and updating such a system would require additional staff practically ruled it out; the White House was under continued pressure to be small and seem small.

But a solution was eventually proposed-- hide the computer in the military. Here a few more staffers and one more computer would hardly be noticed, and the White House could plug into it without actually "having" a computer. The trouble with this idea, as Harden recognized in his capacity as special assistant to the president for information, was that he could not simply request John Carabello, director for data automation in the Office of the Assistant Secretary of Defense (Comptroller), to squirrel away a management data system for the White House. This is not what Congress appropriates funds to the Pentagon for, and Carabello--a professional, not a political appointee--might balk; indeed, even a political aide might ask for instruction in writing. Instead, Harden drafted a brief note which Hamilton Jordan, Carter's chief of staff, sent to Harold Brown, secretary of defense. (If not exactly illegal, this maneuver represented a direct attempt to use funds in a way not authorized by Congress.) Shown a copy of this note, Carabello was willing to work with us, although his conversion to full support came later, during a meeting in the Pentagon itself.

Questions of Access

Carabello saw no particular technical difficulties in setting up the system, although he was worried about the costs, especially the question of staffing it on a continuous basis, and what account to charge for the expenses involved. However, his attitude turned from dutifully cooperative to openly enthusiastic when the issue of terminals came up. The matter arose almost casually as we discussed the details of the system: To "access" the system, we were asked, how many terminals will the White House need? This seemingly simple question opened up the issue of power. Students of the effects of computerization on organizational hierarchies have often pointed out that introducing computers affects the power structure of an organization. Exactly what the effects are is less clear, but there always seem to be some. For instance, computers make it easier for people of various ranks to access a body of information that in pre-computer days was more closely held. These studies have usually dealt with computers used largely to store, manipulate, and retrieve information. A management tracking computer would have even greater impact, since those who had access to it could gain a view of the total performance of any division--theirs or a rival's, even the entire organization's.

I argued that the computer should have one terminal, in the office of Hamilton Jordan, the White House chief of staff. Harden's people, backed up by the Pentagon, could "staff" the service, but only the chief of staff should get the daily or weekly status reports on who was falling behind, who needed backup, who was usually on target. If the computer were to provide an overview, its most appropriate first user would be the chief, who could distribute the information as he deemed fit.

Harden, on the other hand, maintained that there should be scores of terminals within the White House and at least one in each federal agency. He believed dispersal throughout the White House was needed to reduce any sense of threat to other divisions if one division, Harden's, was to manage the information to be fed into the computer and thus in effect become part of the supervisory system. (As Harden saw it, "Anyone who gets ahead of the pack will be eaten up by it.") Also, he believed all the White House divisions could benefit from the information, and of course they would have to feed their performances into the computer to keep it up to date. Moreover, he felt the same way about the government agencies, which I thought did not wish to be closely supervised by the White House and would not be willing to link into such a system. When we visited the Department of Housing and Urban Development, as a trial run, we were told flatly: "We would be happy to join your system as long as our internal agency tracking system could not be accessed from the White House."

What delighted Carabello in all that? As Carabello saw it, the White House was in a "mess," with various divisions each pulling its own way. Working within the military, he was deeply offended by all these squabbles. He seemed to see a great mission to "straighten out the mess" once and for all--via the computer system. It would provide a central focal point for more streamlined management, which "by golly, the place needed."

Resistance to just such central management, reflecting the plurality of viewpoints and power bases within the White House, was what finally prevented computerized tracking from taking off. During this period, Alonzo L. McDonald was appointed to work directly with Jordan on White House matters, and he brought his own ideas of how to proceed with such a system. Hugh Carter did not like Harden's getting more duties and power, and the other divisions seemed to be increasingly suspicious of the whole thing. And, as the support for the tracking system was weak and growing weaker, it was easy to become preoccupied with other matters. Soon, meetings on the tracking system grew less frequent, and staff was gradually assigned elsewhere. The idea seemed to fade away without ever being tested on the White House level. (However, the Reagan administration is reported to be working on one.)

Success--in a Subsystem

Meanwhile, there was an unintended dry run of the tracking system, or at least a fair facsimile of it, in one segment of the White House. The Carter administration's first legislative package on energy, launched in 1977, did not get very far in Congress. On July 15, 1979, after a dramatic review of his general policies, Carter returned from Camp David and addressed the nation, proclaiming that America suffered from a decay of the spirit and that our attempt to come to terms with the energy crisis was to be a measure of our spiritual revival. In the months that followed, the White House pulled out all stops to get the new energy package going. A task force was set up within the White House, with 15 subgroups but with one purpose: to manage the legislative and political drive and the public "outreach" (that is, modification of public opinion) to get the energy package through. The subgroups covered specific topics and their focus cut across federal agencies, congressional committees, and interest groups. They included "solar bank," "energy mobilization board," "windfall profits tax," "Alaskan gas pipeline," among others. While this drive may not have provided a fair test of the American spirit, it surely came to be seen as a test of the administration's ability to get its act together.

Harden and I talked to key staff members on the task force about the advantages of using a computerized system to track and coordinate the scores of efforts that the various subgroups were undertaking simultaneously. Soon a minicomputer and printer were rolled into the New Executive Office Building where the staff was headquartered. Daily, at about 3 P.M., information from all the subgroups was fed into the computer--achievements, benchmarks for next steps ("to do list" and "due dates"), and synergistic acts that required two or more subgroups to coordinate their efforts or required one to wait until another moved. (For instance, a subgroup might have to clear a step to be publicized with the appropriate congressional committees; members of Congress resent reading in the press matters which concern them and about which they were not previously informed.) By 8 A.M. the next day a classified copy of the resulting book was on the desk of each task force member. The system did not perform any miracles--in terms of spiritual restoration or increased energy supplies-- but it contributed handsomely to the managerial effectiveness of this drive.

Admittedly, several factors made the use of a computerized management tracking system easier for the energy task force than for the White House as a whole. First, there was a clear line of authority; there was hence no question to whom the system should report, who should have access to its information, and who would be subject to the commands contained in its daily printouts. Second, the level of reality was easier to deal with. All initiatives were new (or at least treated as such); none already had a reality; all concerned matters that had to be undertaken in a hurry. Third, the matter of the level of abstraction was skirted by treating all subjects as equal, a plausible approach when one deals with 15 subject areas, all concerned with one main topic.

I do not believe it follows, however, that such a system cannot enhance substantially the productivity of top management in a more diffuse, complex, multilevel system. This would include the White House as well as large corporations. But it might require, as it is introduced, some clarification of the lines of authority, a classification (or "code") to separate proclamations and expressions of goodwill from policy commitments, and these from the record of actual accomplishments. And, if the system is not to be overloaded with details, it may well require a separation of information into bits, subcontexts, and contexts. For example, information about unemployment (bits), labor policies (subcontext), and economic policies (context) fit into each other like Chinese nesting boxes and cannot be treated as coequal items.

None of these steps is easy to undertake, from a data-processing viewpoint, and all carry a measure of political hazard of the kind entailed in clarifying the lines of authority and separating useful fictions from facts. In a corporation, for example, the chief executive officer may incur such cost if the tracking computer forces him to state to which vice-president the divisions are to report and on what matters--an issue he may not wish to have cleared up if he sees a benefit in keeping his vice-presidents in sharp contest with each other about duties and authority. In the end, it seems that what determines whether or not such a system will be used is the calculation of the size of the benefits a more productive system promises versus the potential political costs. This is an issue that organizations are still grappling with as they try to apply the power of computers, using them not merely to store and manipulate data but to aid top-level decision makers.

The Communitarian Network
2130 H Street, NW, Suite 703
Washington, DC 20052
202.994.6118
comnet@gwu.edu