Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Managing Microcomputers and End-User Computing Some Critical Issues Roger L. Sisson* Much of U.S. industry needs help. Proper use of data and com- puter power might provide part of the needed assistance. An ex- ample of the kind of help I'm referring to comes from the steel industry, one of those most in need. With access to a computer and the right data, a cost analyst at one company we worked with provided significant information about his company's product profits. This analyst had spent part of his career in the sales depart- ment. Soon after he was assigned to cost analysis duties, he was given access to a terminal, a powerful fourth-generation language system, ant! ciata particularly invoice and cost data. Until this point no one hack been able to explain variations in company profits. As a result of his work in sales, however, the analyst knew that costs and profits were reported only on a procl- uct line basis, and that one line accounted for 60 percent of the company's business. With his computer-aiclec! tools he was able, in a few days, to analyze the profitability of individual products in the major line. This had many benefits: it explained the month-to- month variation in profits; it pinpointed} areas where cost reduc- tions would pay off; and it provided guidance to sales. This analy *Roger L. Sisson is president of Sisson, Michaelis Associates, Inc., Swarthmore, Pennsylvania, and lecturer at the Wharton School of Business. 8~
82 ALlNAGlNG MICROCOAIPUTERS sis will not turn the entire steel industry from loss to profit, but it is helping one company come closer to profitability. Such analysis is becoming routine, an example of the payoff use of computers by nonspecialists. But this use of computers by non-data processing professionals in large organizations raises some important issues. Two issues are critical: the quality of decisions that result when decision analysis is facilitatect by microcomputers or other com- puter-based aids, and the security of ciata and programs from theft or manipulation for unauthorized purposes. There are also less critical, but still important, administrative issues. Before looking at these issues we need to understancI what we mean when we talk about this exploding phenomenon of end-user computing. Computing in this context means accessing and per- forming analyses on ciata to support decision making. The data usually come from existing files but may be entered by the ana- lyst. Computing also covers the case in which the user is installing a small, local, complete system that includes entering and updat- ing data ant! producing stanciarc! reports as well as analytic out- puts. The end user is a non-data processing professional who is per- forming the sort of computing just described. In other words, the main mission of the end user is some function other than data processing. He or she may be performing market research, finan- cial analysis, contract administration, or budget preparation. In considering enc! users and their use of computers in relation- ship to the organization, we may for the moment ignore the differ- ence between terminals and microcomputers as the end user's tool. Whatever the tool, the principal issue is: Does the availabil- ity of computer power facilitate better decisions? Over the last 30 years the data processing profession has devel- oped important standards, guidelines, and procedures designed to facilitate the use of the computer, to make program mainte- nance easier and less expensive, and to insure the integrity and security of data. Do the hundreds of thousands of new end users have to come up with their own sets of stanciards ant! procedures through Tong and sometimes bitter experience? Or can clata pro- cessing professionals help end users? These questions suggest a third important end-user computing issue that, along with qual- ity of decisions and security, deserves attention. This is the issue
SOME CRITICAL ISSUES 83 of administration of information processes ant! resources. Let us Took at each of these issues in some detail. As background! to a discussion of the effect of enct-user comput- ing on decision quality, I would like to introduce the notion of clistributed creativity. This is my name for the much cliscussed concept that organizational success depends on innovation and a striving for excellence at all levels of the organization. In the book In Search of Excellence (Harper and Row, New York: 19823 Peters and Waterman remark that in the excellent firms, "quality and service were invariable hallmarks. To get them, of course, everyone's cooperation is required, not just mighty labors from the top 200" (p. 241. Elsewhere in the book, they offer an example: "3M has been describer! as 'so intent on innovation that its essential atmosphere seems not like that of a large corporation but rather a loose network of laboratories and cubbyholes populated by feverish inventors and dauntless entre- preneurs who let their imaginations fly in all directions' .... They encourage private risk taking, and support good tries" (p. 151. In more philosophical terms, one goal of a large organization is to strike a balance between authority and anarchy. Somewhere between these two extremes lies a freedom to create modified by controls to insure that the organization is cohesive and has direc- tion. The best means of accomplishing this is the major manage- ment theory question of this decade. One way of answering the question goes like this: Wisdom comes from experience and usually correlates with level in the organization. We assume the chief executive officer is wise. Therefore, important decisions should be made at the top. We should centralize the decision making ant! the analytic support. Actual experience, however, has shown that the "top" is too far from the action. There one loses touch with customers, vendors, operation, shops, and labs. To restate Peters and Waterman, the best process is to transmit the wisdom through the culture and to place most of the decision making where the action is. If the cul- ture transmits wise values and constraints, decisions made at lower levels will be based on the best information in a context that guicles the decisions with experience and wisdom of top manage- ment. Enc3-user computing can help both to provicle information and to transmit some aspects of the cultural wisclom. How does it do this?
84 AlANAGING MICROCOMPUTERS Creative leaps of imagination are not inductive. They do not derive logically from the data But creativity does not occur in an environment devoid of data either. The process of studying ciata seems to facilitate innovative thinking. Since the end user often cannot predict which data are relevant, access is needed to a vari- ety of files with data about the environment, the history and cur- rent status of operations, policies and rules, and resource avail- ability. The end user may be performing any of a number of management analysis tasks: looking for exceptions, for opportu- nities, for measures of current status and progress, for alterna- tives, or for estimates of the consequences of adopting an alterna- tive. An innovative insight may come at any point cluring any of these studies. There are of course costs and risks in providing computer re- sources to many end users. Is the cost worth the benefits from improver! decisions? I think the benefits significantly outweigh the risks and costs. This judgment is based on two beliefs: first, that distributed creativity is valuable and is facilitated by end- user computing, and second, that our culture is particularly able to be creative if given the right tools. Especially in the United States, we learn from childhood to do things ourselves. At great national expense we prefer automo- biles to mass transit so that we can go where we want when we want. We have even learned how to drive trucks to move our be- longings ourselves rather than use a trucking service. The ulti- mate demonstration of distributed creativity in our culture is the prevalence of entrepreneurs and the proliferation of small enter- pr~ses. With this cultural heritage it is not surprising that managers and analysts want to handle their own decision support and the computing that goes with it. The rise of personal computing was predictable given our culture of individual initiative, even if most data processing professionals, blinclecl by their own expertise, did not forecast it. But a clesire to do things on one's own is not enough. You can't rent a drive-it-yourself truck if the only trucks available are 18- wheelers requiring special training and hard work to strive. And you won't rent a truck if the closest rental office is several hun- cirecI miles away. Until a couple of years ago computers were like 18-wheelers
SOAR CRITICAL ISSUES 85 they were inaccessible. Communication facilities were not exten- sive; terminals were slow and awkward. Operating systems were unfriendly, making it nearly impossible, even for professionals, to get into the system. And if you did it cost an arm and a leg. Six things have occurred recently to make end-user computing a reality, to make it easy for the user to rent (connect to) the sys- tems, to drive, to react the maps (use software), and to pay for the gas (the computer resources3: · Inexpensive on-line computer resources became available through efficient mainframes and reasonably easy-to-use video terminals or through microcomputers. · Operating systems are now user-friendly. · Fourth- ant! fifth-generation language systems are now avail- able, allowing end users to communicate with computer systems without knowing programming languages. · A large number of people are either knowledgeable about computers or at least not afraid of them. · Decisionmakers have a desire, and usually a real need, to pro- cess information rather than fly by the seat of their pants. · End users recognize that the data they need is available somewhere in the organization's computer files. These breakthroughs mean we can now "do it ourselves." EncI- user computing is here. With 8.3 million terminals and 2.4 million micros in the United States, the 300,000 data processing profes- sionals are outnumbered by end users 33 to 1. Now that we have the tools, will their use result in better deci- sions? Although I have seen no formal studies on changes in the quality of decisions when extensive computer facilities are avail- able to the decisionmaker, there is some evidence. A survey now in progress of encI-user support groups (information centers) sug- gests that end-user computing floes make contributions to the bottom line. Alloway and Quiliard (MIS Quarterly 7~2~:1983, 27- 41) also found, as a result of a survey, that managers certainly want decision-support computer facilities. According to Peters and Waterman, Abernathy, Clark, and Kantrow (Industrial Renaissance. Basic Books, New York: 1983, p. 211, and other management researchers, encouraging innova- tion at all levels is vital to an organization's continuer} success. If
86 M`4NAGING MICROCOMPUTERS this is so, the availability of data-analyzing capabilities and good encI-user computing support should in many cases improve the clecision-making process. There is also a flip side to using computers in decision making. A computer, whether a micro or a mainframe, is a very powerful tool. As with any powerful too] it is possible to use it to make powerful mistakes, inadvertently or fraudulently. One of the po- tentially dangerous mistakes is a bad decision, because it may have far-reaching consequences. Therefore, it is important to know how decisions are being macle by end users, to know that they are being macle in reasonable ways and with the organiza- tion's objectives in mind. This concern was illustrated by an offi- cial of the Bank of America who said to me, "We wonder how all those loan officers are decicling on loans with their IBM PCs and Visicalcs." The freedom of the individual to innovate, given the right tools, is unique in the West ant! most prevalent in the United States. Thus, I believe that the cost of not providing modern information tools to end users is much greater than the cost of a few bac3 deci- sions macle more easily by the use of these powerful tools. Our successful organizations have always depended on their employ- ees to improve them, to innovate, and to advance their welfare. One characteristic of these organizations is that they provide nearly everyone with the tools needed to be creative. A period of increasing international competition is not the time to hoist back. Once we agree that good encI-user computing resources do lead to better decisions, we can go on to consider the difference be- tween encI-user support from terminal-based mainframes and from microcomputers. (Mainframe here means any large com- puter or minicomputer controlled by a data processing staff.) The mainframe/micro choice must be viewed from the benefits side. Which will do more to improve the quality of decision making? I think the choice today depends on the specific system. There are three determining factors: · Access to relevant data may be difficult on a stand-alone mi- cro, but if good communications are available the micro is no cTif- ferent from a terminal. · A mainframe system with very good response time and an array of effective, user-friendly software might provide better support of decision making than a micro. But many mainframe
SOME CRITICAL ISSUES 87 timesharing systems do not meet those objectives. They are slow, and good software for end users does not exist or is not made available. In these cases the ens! user will find the micro prefera- ble. · Mainframes are expensive because of the multiuser overhead and the communications costs. If we recognize that any micro can act as a terminal and do much more, and that micros are not much more expensive than a terminal, microcomputers will probably be the preferred solution in terms of benefits. Three acictitional developments will clinch the micro's acivan- tage over terminals: the capacity of micros is increasing; the friendliness of micro software is improving; and the facilities to transfer data between micro and mainframe, preferably in a way transparent to the user, are becoming available. Soon the micro, combined with communications to the sources of data in the mainframe, will be the dominant solution. (This com- bination of microprocessor and communication, accompanied by good software, is termed by some a "workstation." In addition to these policy and technical considerations, we can- not forget the individuality factor in our culture ant! how it will affect the use of micros. Taking away or refusing to allow some- one to have a micro will be like taking away or refusing someone the use of an automobile. That is unthinkable in a country where we have a hard time preventing even drunks from using auto- mobiles. Encl-user computing, largely supported by micros, is obviously valuable. If there is some reason to slow down or suppress the use of micros it must be because of other issues, such as security or administrative decisions. Security is a critical issue: making a system secure may also limit its flexibility and usefulness. Such limitations prevent or slow down the distribution of capabilities that aid creativity. Therefore, the goal is to make the security procedures easy to use yet relatively impenetrable. A single key is not hard to use. A lock requiring two people and two keys gets to be a nuisance. When I was designing systems I had little concern for safety. The auditors were there to worry about that, and their specific suggestions would be implemented. Security, however, was not allowed to get in the way of ease of operation. Today I think
88 ALlNAGlNG M7CROCOAIPUTERS greater concern is warranted. When thousands of people are ac- cessing sensitive corporate or agency ciata, the probability is that someone will remove competitively sensitive data on a floppy disk. A floppy disk is convenient; it is easy to transfer data to it; it can hold hundreds of pages of data. A person can walk out of a building with it unobserved. The contents neec! not be identifiable externally. How does a security guard know what is on one? There is also the well-publicized problem of illegal access by telephone. Since solutions to this kinc! of unauthorized access are known, the problem falls into the administrative area. It should be possible to detect and trap unauthorized dial-in access. Access to a computer system is possible not only directly through the normal log-on process, but also by tapping communi- cations links ant! by removal of media such as floppy clisks. To eliminate loss through these means some form of encryption is requires! to protect the data even when the meclia are compro- mised, or to make decoding so expensive that potential thieves are dissuaded. These measures are expensive, as are any security measures. To justify them the potential loss from a security breach must be greater than the cost of the security measures. The organizations we have worked with limit their security measures to password systems. Some depend only on the log-on password; others have protection at the file level, so the end user must know two or three passwords. A few have aciclitional password protection at the op- erating systems level. I know of no commercial firm that uses encryption in relation to the kind of end-user computing we are discussing. Ultimately, most organizations depenc! on a trustwor- thy work force. Microcomputers are removed from the direct control of the mainframe, and therefore are not guided by data processing stan- dards. Does the use of micros create adclitional security prob- lems? To answer this we must distinguish between external and internal breaches of security. In terms of external attempts to invade the system, it seems to me that micros do not acid prob- lems. Physically locking the micro and any removable media may be necessary in some cases. However, data clown-loacied to a mi- cro is not at greater risk than ciata printed out, or data put on a floppy disk, or microfiche, or any other removable media. In rela- tion to internal security leaks, the micro, as a powerful informa- -tion processor, may provide opportunities for access and removal
SOARS CRITICAL ISSUES 89 not previously available. Internal security is ultimately still a matter of auditing and personnel policies, not technology. The final set of issues are those that, I believe, can be resolver} in most organizations by normal administrative procedures and proper computer-user interfacing. These issues are the effects of bad data, data integrity, the problem of transferring end-user-de- velopec! programs, and the question of who pays for acicled com- puter capacity, the wasting of resources, and the proliferation of languages. PoorData. Are decisions supported by computer-based analy- sis cleficient because of poor data? This may be an important ques- tion where data from outside sources are used. The quality of out- side data must be checked just as the quality of received parts or materials is checked. (In a few places it may be advisable to check the quality of the internal data as well.3 Data quality really in- volves two questions. Is the decisionmaker better off with no data or with ciata that may have some errors? If flawed! data are user! will the power of the computer-based analytic tools in some way amplify the errors? There are several responses to this problem. Obviously, care should be taken to prevent bad data whether micros, terminals, or ac3 cling machines are being used. In each case the end user is more likely to know the fiats and therefore be better able to catch errors than corporate staff or others. In fact, the power of the computer may help the end user find errors by performing various consis- tency and trend analyses. Thus, end user access to ciata may have an additional benefit of making the data more accurate. Further, the kind of decision-support analysis we are discussing may be insensitive to occasional data errors. Of course, management should always be aware that recommendations ant! analytic results may be based on poor data. End-user computing should not aggravate this situation if administrative procedures are fol- lowed to keep the data as clean as possible and to reminc! the staff to check the data and the reasonableness of the results. Data integrity. Can the end user affect the operational data in corporate files? If so, can the integrity of the fiats be assured? Can the changes made by an end user be audited? The only way to insure the integrity of data is to maintain all the controls that data processing has painfully learned: good edit- ing, effective audit logs and procedures, and good recovery pro
so MANAGING MICROCOMPUTERS cesses. There is no reason to relax these for the benefit of end users. The only way master files should be updated is by the nor- mal, well-controlled transaction processing system. Transfer of data and programs. What about files that a user maintains indepenclent of the mainframe systems? Should the data and programs clevelopecl by an end user be considered pri- vate, scratch lead material or should some mechanism be installed to allow sharing of this material? If the end user is working on a mainframe the data and programs are at least accessible, even if they are not properly structures! ant} clocumentecT. But ciata and programs developed on a micro may be as inaccessible as those on paper. Some organizations are in fact taking the attitude that the data and programs developed by an end user on a micro are the same as material on paper. If the user leaves, the new occupant must develop his or her own routine and ciata. Peers who need to do similar processing must clevelop their own tools, perhaps with the informal help of the first end user. The results of the end user's analyses are submitted in reports (with appropriate appendixes in the usual way. The data and routines not included in the appen- clixes are no longer of interest; they are throw-away materials. If the organization decicles that it does want to capture pro- grams or data that are on a micro and judged to be valuable, acI- ministrative procedures must be instituted just as they are for data processing professionals. These will not be easy to enforce, however, because there are many clecentralized end users who have little interest in data processing's problems. The only other alternative is to prohibit the clevelopment of " systems" on micros and allow such development only in a mainframe context. Resource use and charge-back. If we provide many end users with the opportunity to use computers, and especially if we make it easy for them with effective languages ant! other software, the clemanc! for computer capacity will soar. The issue is whether ad- ditional capacity is a good investment. If end users are helping to make decisions that significantly increase the effectiveness of the organization, the service it provides, or its profits, the cost of the computer resources may be well justified. But only management can justify this resource use. To do this, it should know what the end user is doing and how that work contributes to the organization. The data processing people, who normally have to justify the increase in capacity, do not know the
SOME CRITICAL ISSUES 91 benefits side of the justification. Therefore, charge-back of all computer costs is mandatory, and every company we have worked with or surveyed that supports encI-user computing has a charge-back of real budget clolIars. The budget transfer from charge-back is prima facie evidence for the data processing de- partment that the computer use has value; the customer paid for it. If those in top management clo not agree, they have to discuss it with the end users. Wasted resources. If people are given relatively free access to computers will they waste the resources? Waste can occur in sev- eral ways: · trying to solve problems by random trial and error (with or without models); · computing rather than thinking, researching, or getting out and talking with customers, employees, and other individuals; · using poorly written, inefficient routines; · trying to write programs instead of using packaged software or higher-level languages. Unlike scrap in a metals production process, the scrap in an infor- mation process is invisible. It is not easy to identify waste, so it may become prevalent and expensive. People waste information tools all the time. Who keeps track of pencil and paper wastage? Is a few minutes of computer time spent on an inefficient routine worse than throwing away some sheets of paper from a bad draft? To answer, one has to look at cost and benefit. For example, a little waste in a process that costs a couple hundred dollars but may be saving 10 times that in re- duced costs or improved sales is not worth concern. Choice of languages. Effective end-user computing that leads to better decisions requires languages that end users can handle. A major issue is the selection of the proper languages and sup- porting software systems. This is a technical topic, but it shouIc! not be overlooked as a management issue. It is important also to standardize languages throughout the organization to prevent Tower of Babel. Every group we have contacted has settled on a few basic language tools: for reporting, for statistical analysis, for financial planning, for electronic mail, and sometimes for word processing. (In all cases there was one set of standard software for the mainframe and one set for micros.)
92 MANAGING MICROCOMPUTERS If my thesis about distributed creativity is correct, end-user computing, with micros where justified, is appropriate. Through a good end-user support group, sometimes called an information center, local managers can receive information support of all types, not only with hardware but with software and training. However, some controls shouIc! be attached to these support efforts: . Use passwords properly. Change them frequently. Tie them to people and terminals, not groups. · Control the use of floppy disks and prevent unauthorized re moval of data. · Isolate the user from the operating system with a friendly menu and command-driven front anti. · Don't let end users update main files except through batch processes with good edits and controls or through on-line systems with formal edit, audit, and recovery procedures. · Decide between a throw-away or a share policy for user-devel- oped, decision-aiding programs. [f sharing is chosen, build the in- frastructure to support it. · Provide good user training. Users tend to follow the instruc- tions given, and good training can help reduce waste, correct inef- ficient routines, and promote data security and integrity. · Use a real money charge-back system and let the end users' manager worry about waste. Allocate disk space carefully- it is relatively expensive. Limiting disk space constrains the extent of data and thus indirectly controls how large and complex the us er's systems can be. Distributed creativity is an important national asset. It can help the revitalization of industry and the recovery of the U.S. position in international competition. Enci-user computing, sup- ported by micros or terminals and by a good support staff, pro- motes distributes! creativity. Microcomputers in particular can encourage innovation and good decision making at all levels of the organization. No longer just an executive gadget or a local data processor, micros have become, I believe, an important link in re- building our national strength.
