Click for next page ( 8


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 7
THE OFFICE AS A SYSTEM Michael Hammer Massachusetts Institute of Technology Dr. Licklider, in his opening remarks, has identified a large number of important topics that should be covered today. I have the advantage, being the first speaker, of getting to address all the easy ones and leaving other people to address the difficult ones. What I want to do today is talk about an approach to office automation. Office automation is a buzz phrase, an advertising slogan, that we see in the trade press and the popular press. Office automation has also become a whole new industry and there are a variety of approaches to it. One approach is basically Tom Swift and his electronic office: the idea that you can go into an office, see somebody pounding a keyboard, and by giving them a keyboard with automatic transmission make their work better. This is a view that technological marvels somehow magically make everying better. A view that if we can replace paper with electronics, then ipso facto it must be an improvement. Well, I am doubtful. What I want to talk about today is a rather different approach to viewing offices by approaching office automation with a more systems-oriented perspective. The starting point for this is a set of premises. The first one is really rather bad news. Nobody wants office automation. What I mean by that is described by the following story. A few years ago the president of a well-known handtool manufacturing company was addressing his firm's annual meeting. He got up in front of the room and said, "We had a very good year. Sales were up, growth was up and profits were up." Everybody smiled. Then he said, "I have some very bad news. Nobody wants our drills." The audience was shocked. "What do you mean the customers do not want our drills. They are buying our drills hand over foot." He said, "That is right. Nobody wants our drills. They want holes. We, unfortunately, are not selling holes. We are selling drills and that is not what people want. Drills are the means, holes are the end."

OCR for page 7
We have to keep the same thing in mind in the office environment. Nobody wants office automation. Nobody wants offices. Nobody wants information. Nobody wants the whole information sector of the organization. What people want are cookies, cars and trips to the beach. The rest is just facilitative overhead to help do a better job producing end items. If we keep putting our emphasis on means rather than ends, we will start thinking that the information sector and the office environment is an end unto itself. Then, we will end up in a very bad state. A second premise: people talk about office automation when what they want to talk about is a general administrative office. They say, well what we want to do is focus on the office in general, the office in the abstract, not any particular kind of office, but the office. Well, I have never seen the office in general. I have seen a lot of offices. I have seen insurance offices, sales offices, brokerage offices and government planning offices. I have never seen the office. The platonic idea of the office exists only in the abstract. In any specific instance, there are exact details which are concrete and enormously important. The specifics define that particular office and distinguish it fromn all the rest. The specifics establish what is critical and what contributes to the way the office operates. A third premise of mine is that more and faster are not always better. In fact, they are often worse. There appears to be a general concept underlying most of the conventional approaches to office automation that if you can do more of something, or do it faster, then the result must be better. This concept suggests that if we can generate twice as much text per day then that must, by its own fact, be a better approach. Well, I have my doubts. Twice as much junk is just twice as much junk. If you merely proliferate equipment to produce output — as measured in number of forms, measured in terms of lines of text or number of communications — without trying to focus on what the purpose of the communications are, or what the purpose of the text is, then you are not better off. In fact, you have generated more garbage that other people have to waste their time reading. There are numerous stories about this and they are not apocryphal. They are very concrete. I was recently in a New York City law firm. They had just installed a large number of word processing systems, yet, they had the standard ratio of secretaries to attorneys: namely, one secretary for every two attorneys. I asked the managing parther, "How did you manage before you had word processors? What did you have, lO,OOO secretaries here?" He said, "Oh, no. Before word processing, briefs only had three drafts, now they have forty." So, I said, "Aha. Are forty drafts better than three?" And 8

OCR for page 7
ne said, "I nave no idea." Therefore, I suggest that given the equipment, people will polish prose until the sky falls and there is no measurable way of determining if it is really any better. There is another delightful story, about a government agency that installed a lot of word processing equipment and was delighted to find twenty percent productivity increase by their secretaries. With that twenty percent freed up time the secretaries were giving tours of the word processing center. On the other hand, I also know of organizations that would think anything that enabled the corporate staff to produce more memos faster was a disservice to the rest of the organization. So, merely producing more text can lead to information overload, less productivity and less efficiency, and a high cost of installed capital equipment without any real measurable benefits. There is another effect that I call "the copying machine effect." Some twenty-five years ago, when the process of xerography first came into being, a well-known computer company did a very careful market survey of whether or not they should go into the copying business. They determined that if they captured one hundred percent of the existing dittograph and hectograph market, it would not repay the capital investment required to get into the business. So, they decided not to get into it. Unfortunately a small company named Xerox Halide did not nave the money to afford a capital study and blundered into the business. The point, of course, is that many technologies are catalysts for their own use. While there certainly is a lot of benefit that has come from the use of copying equipment, it did not, at the time, necessarily respond to a felt, perceived and real business need. Having the equipment around merely gave the ability to use it and not always wisely. So, the use of the equipment and the focus on what it is for is key, not just the artifact of what it seems to accomplish. So, let me get back to basics. There are very few offices in the business of just typing or sending messages. Offices and organizations exist to fulfill business or organizational needs and to perform business-oriented functions. I am interpreting business in a very general sense and, of course, I do not mean only the private sector. The business of an office is not to type letters, to attend meetings or to send messages. That is what my three-year old thinks when she comes to my office. She thinks that is what I am doing. I try to disabuse her of that notion. People are really in an office, presumably, to set policy, to conduct analyses and to monitor programs. They have substantive work to do. All the artifacts of the information world are merely symptoms of what is done in order to get the job done. Therefore, if we focus only on the current way we are doing our work without trying to improve or understand the process we may miss the point.

OCR for page 7
What office automation really is is using technology to do a better job, to get the office functions realized in a better way. There are real measures and we will talk about them in a minute. It is not using technology to produce more lines of text per day, or using technology to speed up business communications. Why? What is the matter with business communications today? Sometimes it may be good, sometimes it may be bad. You cannot say that that is an end in itself. If business communications is the rate-limiting step in improving the operation of your office in achieving its function, then you want to speed it up. As a goal in itself, it is pointless. So, with these premises in mind, we can talk about office systems. There are really a number of systems to think about in the office environment. First of all, the office itself is a system. It is comprised of many components: the people, the equipment they use, the information they have access to, the procedures they follow, and most importantly, the mission that they have. If I have a purchasing office, I can say its job is to turn purchase requisitions into purchase orders. The people in the office may change. They may get rid of the old equipment and bring in new ones. Information is up-dated. Procedures may evolve. But, it is still a purchasing office. That is what we have to focus on. We do not want to calcify existing modes of operation by bringing in equipment that instantiates, embodies a particular operation. What we want to do is keep the mission in the front of our mind and design systems to support it. An office is also a component in a larger system; namely, the organization's information sector. The information sector is not a vertical slice of the economy in which a certain number of companies are to be found. It is a latitudinal slice through all parts of the economy. Every organization has an information sector whether it is U.S. Steel, the Department of Commerce or IBM. Every organization has an information sector that is the infrastructure that supports the substantive work that is going on. So, an office information system is the infrastructure inside an organization that supports the performance of substantive office work. The way that an office information system is implemented may vary from case to case. It is not necessarily automated. An office information system may consist of mail carts, typewriters and filing cabinets. That is an office information system. In fact, that is the most common office information system. The automated office information system is the new thing. l0

OCR for page 7
My perspective on office automation is something that I call functional office automation, where we are really focusing on automated office information systems which provide support for office tasks and office procedures in the context of trying to get the office work done. There are varying degrees of automation. There is not a single standard architecture for an office information systems, it varies from case to case. The perspective we have to take is a system perspective and a business perspective. We do not want to go into an organization and focus only on what each person is doing. We want to take a holistic view of the office. We want to say, what is this office about? How can we organize this office to take best advantage of technology and get that job done better. We want to look at it as a system. We want to focus on the application. It is really an application orientation. We are not concerned witrh individual tasks. We are not concerned even with the information processing, but we are concerned with the application that the office is trying to achieve. As Dr. Licklider suggested: the issue is emphasizing integrating components; not putting in a lot of special purpose boxes to handle special purpose tasks, but putting together a lot of pieces, some high technology, some low technology, and organizing them and integrating them so that we can build a system that addresses the application. We also have to have a business perspective, which means that we have to focus on the goals of the organization. What is it there for? We can't be technology driven. We cannot say, "Look, there is some new whiz-bang coming out of the labs. Let us put it into our office and see what it is good for." There is an observable I call the Pirandello effect in office automation. Pirandello was a 20th century playwright. He wrote a play called, Six Characters in Search of an Author. Here we have six solutions in search of a problem. Everybody is tracking the new technology and they feel it is their job to put it into their company. I was talking, about two years ago, to the newly appointed director of office automation for a company in the Fortune lO and he was not quite sure what to do with his job. He decided that his job was to see what new office systems technology was going on in the outside world and try to bring it into his company. I was a little puzzled by that. You don't start with the answer, presumably you start with the problem. You have to look around, see what the needs of your organization are, then look for appropriate technology, then look for solutions, not the other way around. ll

OCR for page 7
Also, you don't want to be task oriented. You don't want to say, "Aha, I see that we make equipment." You should stop to ask yourself, why are we making 9l copies when 9O of them are ending up in the circular file? So, if you focus on what the goals and functions are rather than tasks, that is where we want to go. Furthermore, you should emphasize the uniqueness of the office. Vendors are in the business of focusing on the commonalities of offices. They are interested in large volumes, large market bases. What users need to do is say, what is special about my office, not what is common. If you reduce yourself to the lowest common denominator of all offices, you are not going to get very far. If I can take a piece of generic equipment and plug it in just as easily into a doctor's office or a lawyer's office or an accountant's office or a government office, then it is not doing very much for anybody because those four offices do not have a lot in common. If all I am doing is supporting the intersection of the functions of those four offices I am just taking the bottom base, the smallest piece of what goes on. After awhile things begin to diverge wildly. So, what a user should do is ask what is special about me and how can I use generic equipment in a special way? The goal is not increased efficiency in information handling. The goal is not to produce more letters per day or get messages delivered more quickly unless that is a means to achieving the real end you want, which is to improve the performance of business operations. And, there are measures for those. It might be increased volume with the same head count. It might be faster turn around with fewer errors. There is a whole list that can be developed. The point is that is what you have to focus on and really keep in mind. All too often vendors make sales presentations about how they are going to justify their office systems equipment, by saving you one-third of a secretary. I am not quite sure which third you are going to save. The cost justification arguments that are often very compelling in advance often seem to vanish in the end. You bring in the equipment and people tell you they cannot do things faster, but are doing things more carefully. That is, they are doing things more slowly. So, if you just bring in the equipment to focus on isolated tasks you are not necessarily going to be any better off. When that happens, the vendors quickly turn around and say, well, you do not want to focus on mundane, secretarial cost reduction. You want to focus on the professional staff and improving their productivity. When I hear that I think of what Will Rogers said when told that Calvin Coolidge had died, "How can you tell?" So, when you talk about improving the productivity of professional workers I say, "How can you tell? What is your yardstick, more decisons per forthight?" I am not sure what the effective measure is if you are going to look at what a professional office worker does. l2

OCR for page 7
Lord Kelvin, of course, was a l9th century physicist, rather a dinosaur in his attitudes. Among other things — to reduce his thought to its simplistic form — he said, "If you can't measure it, it is not there." Well, you know what? He was right. If you cannot measure the benefits it is all hot air. So, with that perspective we are ready for the main question, how do you automate an office? The answer is, that is the wrong question. We don't automate offices. What we want to do is improve the operation of the office information system. We are talking about building an automated office information system where we understand an office information system to be a collection of components that support the realization of office functions. So, we want to talk a little bit about what it is, how you build it, and where it fits in your organization. Well, I am going to equivocate right away. There is no such thing as an automated office information system, a general office information system. There is a particular one for every office. We are talking about office-specific systems. The pieces out of which you construct it presumably are going to be moderately generic unless you want to go into the business of designing special purpose chips for you own organization, which I doubt. However, the way that it is put together and some of the software will be specific to your organization. Furthermore, even for a particular office, there isn't a single automated office information system. There are a whole range of possible equipments and functionality. These possibilities include everything from today's office information system of filing cabinets and typewriters to something that is marginally better, such as small expenditures for electronic typewriters with a small amount of memory or major investments such as stand alone word processors, shared logic word processors, electronic mail systems and minicomputer based systems. These are all different possible realizations of different versions of office information systems for the same organization. The key idea to keep in mind is that we are indeed talking about systems, an integrated collection of components, integrated in a special way to better serve the specific office and the organization structure. As Dr. Licklider said, "It may be that you want to have your electronic mail system and your word processing system integrated so that you can prepare messages with one and edit it on the other and send it back with the first one." Even if the pieces do not talk to each other electronically they have to be viewed as being components of a whole. The equipment base is what we build these systems out of. I am not going to get excited about whiz-bangs today. It is whatever makes sense. It could be word processors, minicomputers, local 13

OCR for page 7
networks. It is easy to talk about technology. The point is whatever makes sense with due consideration of functionality and cost. It is a trade-off. You can build very powerful integrated office information systems today if you have three-quarters of a million dollars to spend. You can do a really good job. You buy a main frame computer and a lot of terminals. However, if your office justified that kind of expense you probably would have done it a long time ago. That is what we call data processing. The difference between data processing and office automation to me is really one of degree not one of kind. In the office, we are talking about low transaction volumes, highly interactive work, a place where human factors are really critical. It is essentially the same concept of using computers for an application system. What is important is determining what level of functionality is appropriate for a given level of cost for the eventual pay-off that your organization will achieve. Now, just so I am not entirely vague, some simple functions that I have seen integrated office systems support include the following. Document production is obviously the kind of thing that word processors are used for. This can be either a semi-automated or completely automated production of documents from form letters and similar projects. Information presentation is another useful function. For example, a bank that I am familiar with had a letter of credit system where the person determining if letter of credit terms have been fulfilled calls up the documents and examines them. The system presents total information and enables him to decide if the terms have been supported or not. History tracking and work flow control is another very important support function. What is going on? What is the status of an order? To know where an item is, to know what actions have taken place and to know who is responsible are essential to proper management and can be directly assisted using an integrated office system. There is also a whole set of tools that are generally called personal augmentation tools. These include calendars, electronic message systems, automated note pads, and other similar devices which help people manage themselves and their time. Another category of functional support are the tools for the decision-maker, not tools for the so-called executive, but tools for the person who needs to access data and to use decision support capabilities in order to examine, analyze, project and display information and then make decisions based upon that data. In some cases a real sub-process exists which is amenable to automation and software can be created so the system will do the whole thing.

OCR for page 7
Well, if that is what it is, how do we do it? Again, there is no general prescription, but there are some guidelines. The key issue is understanding the office. That is really the first step. It is not a technological enterprise. The principal thing is understanding what an office is doing, understanding it well, and then seeing how the equipment can help do a better job. This requires detailed first hand observation and analysis of operations, not taking people's word or giving them a word processor. The emphasis of the system has to be on the mission and the function. You want to identify the real problems. It is a Peter Drucker platitude that you should look at problems as opportunities and it also happens to be true. If there is a place where there is a particular bottleneck that is the opportunity to really do something good. The whole thing has to be conducted with a very great sensitivity to the human and organization agenda. Office systems, as opposed to conventional DP systems, are for use by real people, not by programmers. And they have a whole set of concerns that are not to be identified with those of the DP environment. Numerous data processing vendors have badly stubbed their toes thinking, well, gee we have got a text processing system our programmers use to prepare program texts. We will merely put a new nameplate on it and call it a word processing system and put it into an office environment. It does not work. The precept from an office system developer's perspective, the major concept is that rationalization must precede automation. What do I mean by rationalization? I can tell you by a story. There is a story that at the beginning of World War II the British were looking to their coastal defenses and they bought some new truck-mounted guns to replace the horse-drawn ones they had been using for target practice all about the English coast. They were supposed to get real productivity improvements, increased shell firing rates and lower crew sizes. Those improvements were not materializing. They brought in a time and motion man who watched what was going on. He took slow motion photographs. He brought out his stopwatch. He found that the time from when the muzzle was loaded until the time when the muzzle was unloaded, that whole period, two of the five-man team were standing at attention doing nothing. He asked them, "Why are you doing that?" They said, "That is what procedure says. That is what it says in the book." They did not know why they were doing it. So, he went around asking everybody why they were doing it. Nobody could figure it out until he came to an old Colonel of Artillery. He said, "Oh, yes, they are holding the horses." l5

OCR for page 7
So, automating a mess gives you an automated mess of questionable utility. What we are really talking about is not office automation but office design. The design of office procedures, office jobs, office structures and then using technology almost as a catalyst. I have seen cases when it was said, we are going to have an office automation effort and everybody gets very excited and starts analyzing their office and figuring out what is going on. By the time they have rationalized and redesigned their office they do not need the equipment. The equipment gives them the last twenty percent of the benefit. The first eighty percent came from the fact that they have been given the impetus to study and redesign and analyze some archaic procedures. However, you cannot willy-nilly walk into an office and expect everybody to be rational. You have to face the fact that you are perturbing an existing system and that office systems are delicate and have evolved over time. People may say that you cannot change anything. Hogwash, you can change everything if people are on your side. If you give people incentives that are high enough they will stand on their head and spit nickels. So, the incentives are what are key, not improvement to the organization as a whole. You cannot say, well, you are doing it for the greater glory of the Department of Commerce. That will not go over so big. But, if there are some particular advantages for the person involved, they will do almost everything. You also have to face the fact that you cannot always get there from here. We may have an office that operates a certain way and in the abstract what we would like to do is to chuck it all out the door and start from scratch. That is not going to happen. You also have to take into account that there are limits to rationality, that pure systems thinking does not carry all the way. You might walk into an organization and say, this doesn't make any sense. The guy in the corner there isn't doing anything. You don't neeu him. Well, it turns out he is the boss's brother-in-law and you really do need him. So, in designing systems and thinking about human factors — and I mean that in a very broad sense — people are the most important part of the system. If you get carried away with technology and analysis you can run into a lot of trouble. Currently, there is a lot in the literature about what it is going to be like to work in an automated office. There is one point of view, put out principally by people trying to organize office workers and scare them, saying. "It is bad news. It is going to be very much like working at drill presses and lathes. It is going to be an industrial model at the office, except you are going to be sitting at a cathode ray tube all day and frying your eyeballs." l6

OCR for page 7
The optimists, who tend to be vendors trying to hawk their goods say, "Oh no, on the contrary, it is going to be a paradise on earth. We will have no more boring work. Everybody will sit and think all day." Well, in fact it can be both or neither or somewhere in between. It all depends on how you approach it. It all depends on how the system is designed. There are a whole lot of concerns, one of which is human engineering of equipment, but that is only one. Namely, is it something that people can use without trying? I have seen numerous people trying to use poorly designed equipment and literally crying because the system when they did something wrong came back with the comment that said, "Slash, slash you, stack overflow, all jobs and files deleted." It did not work very well. Or, equipment where the inset and the delete functions were on the same key, one in lower case and one in upper case. This is true, not a joke. Needless to say human engineering is very important. However, it is not the only thing. We have to talk about what are people going to do? You have somebody who is sitting there reasonably contented, doing a certain kind of job and you say, "All right, you have a new job." "Me? I didn't ask for a new job. What do you mean changing my job?" So, you have to take into account the fact that you are changing people's jobs. By designing systems you are designing jobs. You have to face the fact that you have to introduce the system. That is often a very difficult thing and you may have to overcome a certain resistance to change. The resistance is not to automation, but to change. In fact, often people like equipment. It gives them something to talk about. But, they are afraid of uncertainty and they are afraid of change that does not carry any benefits for them. There is also a key issue of skill degradation or enhancement. Are you going to turn people today who have a variety of intersting jobs into essentially an entry person staring at a key board all day typing in data that comes in from other sources, or are you going to remove the boring work and turn them more into paraprofessionals who have a chance to exercise some autonomy and responsibility? There also is the whole question of monitoring and privacy in that context as well. Are you going to give people additional responsibility? At the same time, are you going to give them the tools to do it? Are you going to use the new systems technology to better keep track of what people are doing in a way that they feel to be intrusive on their own legitimate needs for privacy? You have to take into account that there is a social framework in the office. There is a well-known case where a certain organization put in a highly expensive word processing center. It was a big capital investment. They decided it had to be used all day, 17

OCR for page 7
eight hours a day through the main shift. Well, they got little or no cooperation. The system fell apart. Why? Because the operators realised that if it was going to be done that way they could not take their lunch breaks and their coffee breaks together and they refused to cooperate. There is a whole issue of career paths. What futures do people have? Are you condemning them to an endless life in front of the terminal, or are you saying, this is an upwards step in a variety of ways. There is a whole issue of how do people view office work and themselves. You can get pretty philosophical here, but what is the work about? Are they becoming slaves to the machine or is the machine becoming an adjunct to them? As I said before, a key issue is really one of incentives. We talk about an OIS, an office information system, as a system, and it therefore has a life cycle like any system. Let me just quickly identify the major points. There is the issue of getting it in the door. There is the problem of high cost. You go to management and say, "We can do a better job if you give us $2 million for our office." And management will say, "You come back with a 3OO page study proving it and maybe I will believe you." High entry costs require a lot of justification which is often difficult to do in advance, because you do not really anticipate what the real advantages are going to be. So, low cost initiation and growth is much preferred to a monlithic, mammoth, all-at-once approach. There is an issue of getting it in and getting it institutionalized, so that it is no longer an experiment, no longer an oddity, but part of the way the system really works. Once it is in, it is going to start changing. It has got to be enhanced and expanded. You want to increase functionality. Right. We brought it in to produce our documents, now we also want to keep track of our work. Also, you are going to grow in volume. Yesterday we had ten people working here, now we have twenty-five. Help! Then there is maintenance. Now, maintenance in this environment is a little different because we are basically talking about software kinds of systems. Cynics, of course, describe software maintenance as replacing old bugs by new ones, but we will not say that. Wnat it does mean, of course, is maintaining things in the face of change. It does not mean that you have a lot of friction, moving parts, and you have to change the oil every 5O,OOO key strokes; but, rather that the technology changes. You want to bring in new equipment and there is an organization application change. You want to do new things and the old equipir.cr4 has 'o . 18

OCR for page 7
The implication of this is that you don't want to get caught up in focusing on the technology. You do not want to focus on the technological realization, but rather on the logical function. If there are different implementations of that over time, that is terrific. As an implementation strategy I find that it is useful to get going quickly, get something small up fast. If it is small and not too expensive, people do not get too excited. You can often justify a small start out of some backhand pocket and you do not have to go through twenty-five levels of management to spend a lot of money. Once it is in, and you start to prove its worth, then it becomes easier to grow and expand. The low-key effort requires less initial effort and limits expectations. It also limits trauma. Not everybody gets exciteu. You can bring it in without ruffling too many feathers. If you are going to bring in systems you have to face the fact that you have to have diversity and flexibility. You should not say, "Right, this is the way we are going to do things here." But, rather you have to have an environment in which people can, within tolerable limits, explore their own approaches. You need systems that are modular rather than monolithic, so they can be introduced over a period of time. In order to do this you have to have a technical warning system, a picture of where things are going. Let me start to wrap up. Where does an automated office information system fit? I see two levels. First, there is the specific office information system that is under local autonomy and control. The whole point is that if I have a responsibility for getting a job done, it is, to put it mildly, crazy to say that I do not have the authority to acquire and manage and use the equipment I need to get my job done. I should not have to go hat in hand to somebody with the name of information resources manager or office automations czar and ask him to solve my problem for me. It is my problem, why am I asking him to solve it? Because he has this certain expertise. Well, he should rent it to me or loan it to me. But, if I am the one with the responsibility to get the job done, then it is my responsibility to have the equipment to help me do it. And, in fact, conventional approaches to centralized, large-scale, corporate systems are really an artifact of the fact that we were dealing with expensive things that had to be shared among a lot of people, which could not be cost justified on a local basis and that required a lot of resource ana expertise to operate. Those factors are changing. The concept of local autonomy and control is the thing that is motivating distributed data processing and is the reason why people are putting in minicomputers to support local operation. The rationale of local control and use equally applies to local office infomation systems. So, this is not an organizational control system, it is a local system controlled by the operational manager. l9

OCR for page 7
However, the infrastructure, the corporate networks, the data communications and telecommunications networks, the corporate data bases, the things that transcend individual office needs, should be managed by some centralized staff. This is the second level. The shared, more expensive resources and the interconnection of communications and integration of individual systems. The systems belong to end-users, but the integration infrastructure belongs to some centralized staff. System integration and the connection of communications are key problems that Dr. Licklider alluded to. Namely, how do you put pieces together? How do you avoid a distributed system with dispersed responsibility from decaying into anarchy? For example, I build a system and you build a system. A year down the road we say, gee, you know it would be very good if my system could send documents to your system or that your system could get data from my system. We find that they cannot talk to one another. That is not so good. But how do you achieve • interoperability without putting an iron hand on control which stifles innovation and creativity? The answer tends to be singular, by means of standards. What we want to achieve is coherence through interface standards, that this infrastructure that ties the peices together is the thing that really enables us to define standards for everything to talk without imposing internal standards on what they have to look like. Let me conclude then just by saying some obvious platitudes and one last story. I told you at the beginning that the bad news was that nobody wants office automation. Well, I would like to conclude with some good news. I am not quite sure what it is, but there is an old story that once there was a king who heard that there was a famous wise man in his kingdom. So, he had him brought to the palace and said, "I hear you are very smart. If you are smart, I want you to teach my horse to sing." The wise man said, "Oh, that is a pretty hard job." And the king said, "I know it is hard. If you can do it, I will give you a bag of gold and if you can't, off with your head." So, the wise man said, "I need some time." And the king said, "Fine, you have a year." So the wise man took his horse home and he told his family about the story. They started to cry and wail. He said, "Calm down, don't get so excited. A lot can happen in a year. Maybe the king will die. Maybe the horse will die. Or maybe the horse will sing." So, there are a lot of possibilities, but you have to be optimistic. 20