Emerging Technologies in Work Design
Paul A. Attewell, Beverly M. Huey, Neville P. Moray, and Penelope M. Sanderson
The last 15 years have seen dramatic changes in the competitive situation of many sectors of the U.S. economy and also in the levels of investment in and deployment of new technologies. These developments present new challenges for human factors specialists. They raise issues about how to design jobs to make sure that the latest technologies fulfill their promise of raising industrial productivity and competitiveness. In addition, they stimulate conceptual questioning about the proper relationship of humans, machines, and systems. Experiences, both good and bad, with emerging technologies have stimulated new philosophies of design, have highlighted gaps in our research knowledge, and have suggested new avenues for human factors research. In this chapter we examine the implications of recent technological and economic changes for job design and for research in the human factors community.
The Economic And Technological Context
Over the last two decades, the world economy has been rapidly shifting from a structure in which a few geographic centers dominated world trade to a more complex, multicentered structure. The rapid expansion of world trade has been accompanied by the emergence of new industrial powers and the erosion of dominance of older centers (National Research Council, 1990).
In the United States, this was experienced, during the 1980s, as a crisis of competitiveness (President's Commission on Industrial Competitiveness, 1985). In many industries, U.S. firms lost substantial domestic market shares to foreign imports, and a similar loss occurred in export markets.
Industrial technology is viewed by many as critical for success in this environment of intensified competition, and a backwardness in utilizing advanced technologies in manufacturing has been identified as one cause of the earlier U.S. decline (Jaikumar, 1986; National Academy of Engineering, 1988; Adler, 1991). Out of these concerns came several policy recommendations for U.S. firms: among these were that U.S. firms make changes in traditional product development processes, accelerate the implementation of new technologies, adopt strategies of continuous improvement, and embrace practices that encourage employee involvement. More recently, the federal government has embraced total quality management (TQM), not only as a strategy for itself but also as a requirement for its industrial suppliers.
The recommendations and diagnoses of several commissions emerged just as many U.S. firms were implementing a host of interlinked organizational changes—from outsourcing to de-layering to using high-performance work teams—while making substantial technological investment, most notably in computers, communications, and robotic technologies. The implementation of this mix of technological and organizational change was far from smooth. Although there are examples of increased productivity, there are also troubling indications that some technology investments have failed to boost productivity to the extent expected (Adler, 1991; Attewell, 1994; Computer Science and Telecommunications Board, 1994).
The problems of design, implementation, and operation of new technologies have therefore moved to center stage. Human factors specialists are faced with technologies that are complex in purely engineering terms but whose effectiveness also appears to depend on teamwork and a host of other organizational innovations (National Research Council, 1986; Adler, 1991; National Academy of Engineering/National Research Council, 1991). Job designers therefore face a dual task: to design jobs to take advantage of new technologies while fitting these jobs into new organizational structures and strategies.
THE CHANGING GOALS OF JOB DESIGN
Early in this century Frederick Taylor (1911) and the Gilbreths ( 1973) were the first to argue that job design should be undertaken in a systematic manner, drawing upon scientific knowledge of human motor and cognitive capacities, of fatigue, and of attention span. Later, the sociotechnical systems perspective added the idea that effective work design
must take into account the interactions among individuals in the workplace as well as the interplay of human and machine.
Drawing upon these insights, various methodologies have developed for analyzing jobs to discover their constituent tasks, and rules of best practice have emerged about how to combine tasks most effectively, where to draw boundaries between jobs, and how to join jobs together.
However, the principles and practices of job design have not remained constant over the last half century, in part because the goals of design have shifted. Different design philosophies or theories focus on optimizing different aspects of work. Over time, design philosophies have shifted as different goals or dimensions of work rise and fall in importance. This point can best be illustrated by a brief historical review of the major approaches to job design and a consideration of some new goals of work design that have emerged or have at least become more salient in recent years.
Taylorism or scientific management dominated job design prior to World War II. As a theory of design, it was centrally concerned with the optimization of physical effort in order to increase speed of production. Its various offshoots, such as time and motion study, sought to eliminate superfluous movement through design of both the job and its attendant machinery (jigs, machine tools, feeding mechanisms, etc.). But Taylorist attempts to optimize on physical effort and speed had important consequences for other aspects of the job:
There was a separation of ''indirect labor" (planning, preparation, maintenance, quality control) from direct labor.
Jobs were subdivided as far as possible, so that each job encompassed a narrow range of repetitive tasks. This was intended to enhance work speed and reduce the time needed to learn the job.
Worker discretion was as far as possible eliminated in favor of "one best way" (the designer's way) to accomplish the tasks (Knights et al., 1985).
The introduction of the assembly line, intended to optimize the efficiency and flow of production, led to further changes in job design. Jobs were designed around the need for continuous (uninterrupted) production. Machine tools became highly specialized, and feed and conveying operations were automated whenever possible. Assembly line jobs encompassed only a few repetitive tasks requiring minimal discretion or knowledge. The emphasis was on speed, simplicity, and stamina. The resulting work design philosophy, an extension of Taylorism, is often called "Fordism."
Taylorism and Fordism were fairly successful as theories of work design, in terms of their professed goals. But they eventually came under
great criticism because they unintentionally degraded important aspects of work and, therefore, undermined other goals of production. From the 1950s onward, sociologists demonstrated that Taylorized assembly line jobs were boring yet physically demanding, leading to stress, low worker motivation, frequent absenteeism, and labor conflict (Walker and Guest, 1952; Kornhauser, 1965). Ultimately, this led to the emergence of a new work design movement, the human relations approach (McGregor, 1960), which stressed a very different goal for job design—improving job satisfaction or psychological fulfillment as a means of increasing workers' motivation.
This goal implied very different design principles:
To encourage meaningful work and a sense of achievement, multiple tasks should be grouped into a single job in such a way that an individual could gain a sense of completion.
Task variety and job rotation are desirable.
Some choice over the sequencing of work, methods, and speed should be left to the employee.
The human relations approach to work design expanded from the 1950s on and became linked to a movement within job design known as the "quality of work life" movement. The approach remains especially salient where high levels of worker motivation (high initiative, good judgment, and carefulness) are important requirements of production processes.
The very success of Fordism in maintaining the high speed of the assembly line led, in many cases, to chronic problems with the quality of products. Alienated automobile assembly line workers who were required to work at a fast pace had a tendency to allow poorly constructed cars to go forward even when there were obvious flaws. Workers would take shortcuts that enabled them to keep a fast pace but that could undermine the quality of the product.
With intensified international competition from better-quality imported goods in the 1980s, the inferior quality of some U.S.-made goods became a major worry. Attaining and maintaining high standards of quality consequently became a central preoccupation in many U.S. manufacturing firms (Deming, 1986). This in turn has influenced job design. Quality testing has been reintegrated into many machine operators' and assembly line workers' jobs: instead of a separate quality-testing staff at the end of the production process, assembly line and other workers have been encouraged to "build quality in" and prevent bad work from going forward. There has been a renewed emphasis on job rotation (even during a single day) and on paying
workers for developing maintenance and repair skills. Some production jobs have been broadened to include cleaning one's machine and surrounding areas, under the logic that this increases a sense of ownership and pride in one's work.
A commitment to high-quality products places new or increased performance demands on production employees. There is a greater need for relatively intangible skills such as taking responsibility and showing good judgment: for example, many of today's assembly line employees are empowered to pull the "andon cord" that brings the assembly line to a halt if they see a quality problem. This was not the case in Taylorist or Fordist job designs. Other new skills are intellectual, such as the substantial statistical and conceptual skills expected of workers who participate in quality circles. Cole (1979) delineates the mathematical knowledge required for statistical control procedures, including familiarity with elementary probability theory, statistical variance, and quasi-experimental approaches to diagnosing sources of error.
The growth of the total quality management movement has resulted in an increased respect for the intellectual contributions of shop-floor and office workers. In many workplaces they are encouraged to join quality circles to look for ways to improve production processes and to save resources. They may also be trained in group problem-solving techniques and in teamwork more generally.
In addition to changing the work lives of many employees, TQM's emphasis on quality has highlighted certain gaps in the human factors' scientific knowledge base. One important kind of quality flaw stems from operator error. Yet we know surprisingly little about the causes and sources of human error and even less about designing jobs and processes to reduce the frequency and seriousness of errors. Research is also needed to integrate the knowledge base on human error into design principles for machinery, displays, work procedures, and communications processes. This is one of several challenges that emerging technologies and the current economic climate pose for human factors research.
Workplace Health and Safety
Another equally important goal of work design, made more prominent by recent technological developments, is the maintenance and improvement of workplace health and safety. Recent concerns over workplace safety and occupational health reflect Occupational Safety and Health Administration legislation in the United States and, in Scandinavia and elsewhere in Europe, the influence of trade unions, which have targeted workplace health as a major issue (Butera and Thurman, 1990). Employees are demanding safer workplaces; legislation has declared that to be their right and has placed an
obligation upon employers to provide safe jobs. In the United States, burgeoning health insurance and disability costs related to workplace illnesses provide an additional powerful impetus for employers to attend to these problems; this will ultimately result in demands for safer, healthier workplaces via better-engineered and designed work (see Chapter 4).
Designing jobs and work processes with health and safety in mind builds on traditional ergonomic concerns. Machine guards, good lighting and air quality, and concerns about protection from noise, lifting and stretching, fatigue and attention span, and legibility of signs and displays are all long-standing human factors issues. They will remain basic to design for tomorrow's workplaces. But the range of factors to be considered in work design will probably need to be broadened to include other, more social and psychological stressors in the workplace.
Consider, for example, the recent upswing in musculoskeletal disorders found among employees who work at computer terminals. In certain U.S. industries, such as telephone companies and newspaper offices, epidemiologists have documented recent mini-epidemics of both wrist pain and neck and upper-back problems, so-called "repetitive strain disorders." Such complaints may affect from 20 to 40 percent of workers in certain workplaces. The problems span a spectrum from severe conditions—such as carpal tunnel syndrome, which can be so painful that sufferers are unable to work and therefore often resort to surgery—to less dramatic, but still consequential, intermittent pain, which can affect workplace productivity and employee morale.
The traditional ergonomic response to such repetitive strain disorders is to study and reengineer workplace machinery, utilizing our knowledge about vision, posture, hand motion, and other areas. Thus, we have seen the design of better chairs, video display terminals, lighting, and keyboards, drawing upon the human factors knowledge base.
However, while giving traditional ergonomic factors their due, the epidemiological research suggests that additional sociopsychological factors may be at work to create job stress that is in turn related to the physical symptoms of repetitive strain disorders. These disorders can be found even in offices that have invested in ergonomically sound equipment. Epidemiologists at the National Institute of Occupational Safety and Health and elsewhere have documented that a variety of job stressors—from fear of job loss to the variability of work, from the existence of deadlines to the nature of supervision and surveillance—are related to these disorders in certain workplaces. It will take considerable design ingenuity to minimize such stressors.
Current design responses to repetitive strain disorders include enhancing work-group dynamics and scheduling rest breaks and job rotation, as well as designing better keyboards, monitors, and other equipment. Unfortunately,
there is little research that examines whether these responses are effective in countering the spread of repetitive strain disorders. Beyond studying the efficacy of these particular interventions, there is a need for human factors specialists to orient research toward the intersection of ergonomic knowledge, sociotechnical factors, and epidemiology with the goal of understanding the links between sociotechnical design, workplace stress, and worker health.
Intellectual Work and Teamwork
For several decades the proportion of the labor force in white-collar jobs has been growing, with managerial, professional, and technical occupations leading the way. As "knowledge work" has come to dominate the information economy, job design has had to embrace new concepts and approaches especially suited to intellectual and informational jobs. Because knowledge workers work at machines, most notably computers, traditional ergonomic concerns with seating, keyboards, and displays remain important. Much attention has also been focused on such human-computer interfaces as displays of information and cognitive maps. But beyond this, computing technologies have changed—and continue to change—the division of labor and the content of jobs, often in dramatic ways.
Before computerization, most information jobs in large firms were subject to Taylorist design. In insurance offices, for example, paper flowed from department to department in a sequence: first to the claims examination department, then to the accounts payable department, then to the check-writing department, then to bookkeeping. Within each department, tasks were narrowly subdivided, routinized, and linked in a chain. Each department typically kept its own databases and paper files. Thus, a single business process, for example accounts payable or assessing an insurance claim, involved sending the originating paper through a long multistep multi-department chain, each link of which consisted of a specialized clerical worker performing a narrow repetitive task. There tended to be backlogs of work in each department, and each department expected a substantial period of time (e.g., one week) to clear a claim through its part of the process.
Taylorist design could efficiently process very high volumes of business items, but was slow because of the many handoffs from person to person. A given claim or payment often took weeks to process from start to finish, and there was no easy way to locate a particular piece once it had entered the stream.
Interactive or real-time computing was widely adopted from the early 1970s on, and it made this fine division of labor obsolete. Real-time computing
allowed previously distinct databases to be linked. As soon as a piece of information was entered, all the relevant databases were instantly updated to reflect the new datum. A further consequence was that previously separate clerical or informational tasks could now be integrated. So, when an order arrived, a clerk could call up on the computer a database that would first determine the customer's credit, that is, whether any bills were outstanding. If the buyer was found credit worthy, the software would present inventory data to see whether the requested item was in stock. If sufficient stock was available the clerk could have a "pick slip" printed at the warehouse giving the order to ship the goods. At the same time the accounts receivable database would note that the customer should be charged, and an invoice would be prepared. And the company's books would be updated to note the additional amount receivable. If the items were not in stock, the clerk could initiate an order for the shop floor to produce the items and could even (via materials requirements planning, MRP software) set in motion orders for more raw material needed before production could commence.
In sum, the interlinked databases allowed for a reintegration of previously fragmented tasks, each of which was previously a separate job in a long clerical sequence. Instead of simplifying steps, the new design logic was to give each order-taking clerk access to all of the steps and activities needed to complete an order, to make a job coincide with the complete range of informational tasks associated with an order. New complementarities also arose between data retrieval, customer service, and data entry. Because databases were now easily accessible, it made sense to give computer clerks the role of intermediary between the information system and the customer. So a customer could phone and ask the clerk whether certain goods were available for shipment, and if not, when they could be produced and delivered. At the same time, the clerk could enter data on the specfics of a new order, payment method, and shipping details. Thus, one-time highly specialized and routinized clerks became more multifaceted "customer service representatives" (Attewell, 1992:71-74).
At first, job designs like these evolved spontaneously. But experiences with interactive computer systems have become codified and today inform a new practice of job design known as "business process reengineering" or (more grandly) "reengineering the corporation" (Hammer and Champy, 1993). This approach looks for previously fragmented tasks to integrate; it "delinearizes" and resequences tasks, creates new, broader jobs, and also restructures (typically reduces) managerial controls to fit the new, broader division of labor. Proponents of the approach claim that it increases productivity, improves service, and enhances job satisfaction.
While new information technologies have created many of these multifaceted clerical positions at the lower end of information work, they have
also stimulated change in higher-paid information work. Today's managers use management information systems—streams of production and financial data—to survey and understand their areas of jurisdiction. This in turn can lead to rather different styles of managerial control and decision making than those found a generation ago (Attewell, 1992), styles that are heavily dependent on quantitative data and computerized decision tools. This can have both positive and negative consequences. Studies of the use of decision tools such as spreadsheet models (e.g., Kotteman and Remus 1987; Kotteman et al., in press) indicate that managers can become dependent on the use of these techniques even when they fail to produce better (or even adequate) decisions. In particular, the researchers cited above have found that managers systematically overestimate the effectiveness of these methods. They have coined the phrases cognitive conceit and the illusion of control to describe these effects.
Such studies indicate an emerging trend in human factors research to examine trust in machines (and in computers and data) as well as the emotional and cognitive consequences of dependence on cognitive tools such as computers. (This is clearly related to the issue of human error discussed earlier.) We shall return below to ways these concerns have become reflected in job design and philosophies of appropriate levels of automation.
A rather different aspect of the impact of computer technologies on jobs involves the rising importance of teamwork in modern workplaces. Teamwork among small groups of skilled individuals predates the diffusion of information technology: one thinks of physican-nurse teams in a hospital operating room, the flight team in an aircraft cockpit, and the team of sailors taking bearings and steering a ship into port. However, several commentators have argued that small-group teamwork of this kind has become increasingly widespread in the modern workplace, going beyond the traditional professions and into new work domains. Frequently, the tasks involved in knowledge work in modern workplaces are highly complex and therefore benefit from coordinated teams of highly skilled individuals, often with different specialties, working together (Peters and Waterman, 1983; Savage, 1990; Drucker 1993:83-89; Mills, 1991; see also Hill, 1992).
This increasing emphasis on teamwork presents a challenge to human factors researchers and the behavioral science community generally. First, there are lacunae in our basic knowledge of small-group dynamics. As a recent review (Simpson and Wood, 1993) put it: "Despite the widely recognized importance of groups, basic social processes underlying group dynamics have received scant and intermittent attention. This has been particularly
true within social psychology." (However, see McGrath, 1984, and Hackman, 1987, for a more optimistic view.)
Second, the research literature on work groups in situ—in real work settings as distinct from laboratory experiments or simulated workgroups—is sparse. There are, however, notable exceptions, for example, R. Helmreich's research on flight cockpit crew dynamics and on operating room surgical teams.
The potential importance (and practical difficulties) of applying human factors approaches to the job design of group work or teamwork can be illustrated by the example of software development, a steadily growing area of employment in our postindustrial economy. Most software development projects are carried out by small teams of systems analysts, programmers, and coders. The work is creative, and for that reason uncertain: up to 25 percent of projects fail, and cost overruns and missing deadlines are endemic (DeMarco and Lister, 1987; see also Computer Science and Telecommunications Board, 1990).
Given the high cost of software development projects, great efforts have been made to increase the productivity of programming efforts. Metrics of productivity and output have been developed (Boehm, 1987). New programming languages and software tools abound; it is claimed that many of these increase productivity. Unfortunately, differences in programming technology do not explain observed differences in software engineering productivity (DeMarco and Lister, 1987), and the introduction of new programming tools has not solved problems with software productivity: several studies have found that the most modern of Computer Aided Software Engineering (CASE) tools are associated with lower productivity (Banker et al., 1991; Orlikowsky, 1988.) Thus, the attention of researchers and job designers has been drawn to sociotechnical factors, most notably the processes of coordination and division of tasks among team members.
Perhaps because of the incompleteness of scientific research, one finds dramatically different approaches to job design in these settings. Some authorities, viewing coordination as the Achilles' heel of software teams, pursue design strategies aimed at removing ambiguity and complexity from teamwork. Tasks are decomposed, simplified whenever possible, and organized as modules to reduce the necessity of coordination or interaction among team members. Lines of authority are made as specific as possible, and standard operating procedures are spelled out. Meetings are formalized and scheduled regularly (Boehm, 1987; Brooks, 1987), and standardized methods of documentation and testing are prescribed. This approach has come to be known as "structured development" or simply as "programming methodologies."
One difficulty with this approach is that by creating formal structures to
minimize the burden of coordination and communication among team members, designers may reduce necessary or fruitful communication, resulting in inferior performance. Kiesler et al. (1994:226) argue that the most productive combination is moderate levels of formalization and structure combined with moderate levels of team communications. At either extreme—high structure or high communication—team productivity suffers.
In contrast, those who take an ethnographically informed approach to job design for software teams are quite hostile to using formalized methodologies to structure team dynamics and programming. DeMarco and Lister (1987:114) give the following critique:
A Methodology is a general systems theory of how a whole class of thought-intensive work ought to be conducted. It comes in the form of a fat book that specifies exactly what steps to take at any time, regardless of who is doing the work, regardless of where or when. … Methodology is an attempt to centralize thinking. All meaningful decisions are made by the Methodology builders, not by the staff assigned to do the work. … They do this by trying to force the work into a fixed mold that guarantees:
a morass of paperwork,
a paucity of methods,
an absence of responsibility, and
a general loss of motivation.
DeMarco and Lister (1987) turn elsewhere for factors that influence effective teamwork. They argue that software productivity requires deep thinking as well as communication. They find that modern workplaces rarely provide good environments for thinking, especially for "flow," a kind of thinking requiring extended periods of concentration and focus. In many workplaces there are frequent interruptions from phone calls and colleagues, and noise levels are high—phenomena inimical to flow. DeMarco and Lister find, for example, that being able to silence one's phone makes a substantial difference in productivity. Teams do better if they have receptionists who shield them from interruption. DeMarco and Lister make a series of suggestions regarding the design of work spaces, noise reduction, recruitment of team members, and style of leadership, which include traditional human factors concerns as well as sociotechnical factors.
Software development is but one of the numerous contexts in which teamwork is the preferred strategy for organizing work. But the contrasting prescriptions and beliefs about how to best design teamwork prove that the knowledge base and science behind job design of teamwork are at very rudimentary stages. This is an area of great promise and great practical importance, but one in which relatively little has yet been accomplished.
As information technologies become increasingly capable of performing a wide range of functions in both manufacturing and white-collar work, attention has turned to the proper role of humans in the system. There are two extreme design philosophies (Kantowitz and Sorkin, 1987): (1) the human should be eliminated entirely from systems, or if that is not possible, the human role should be minimized (i.e., automate as much as is possible) and (2) the human operator should be involved as much as possible in the system, even if artificial tasks must be created to accomplish this. The first philosophy, which is technology-driven, is defined by decisions about when and how to automate (Air Force Studies Board, 1982), which, in turn, define the role of the operator in the system.
Researchers have reported that system designers often believe that the greater the degree of automation the better (Bainbridge, 1982; Boehm-Davis et al., 1983; Wiener, 1985; Wiener and Curry, 1980). However, this moves the human further from direct contact with the system; this has been found to sometimes result in negative consequences when the operator is required to intervene (Air Force Studies Board, 1982). New types of errors or accidents have often been created because automation has changed the nature of the human-machine relationship in unforeseen ways (e.g., Hirschhorn, 1984; Moray and Huey, 1988; Nobel, 1984). A workshop on automation (Boehm-Davis et al., 1983) identified five problems associated with highly automated systems: (1) newly automated systems do not usually provide all the anticipated benefits; (2) when automated equipment fails, the system loses credibility; (3) automation usually increases training requirements; (4) designers often fail to anticipate new problems that will be created by the automated systems; and (5) automation changes the role of the human from active controller to system monitor. Numerous other researchers have also identified both pros and cons of automation (Bainbridge, 1982; Wiener, 1985; Wiener and Curry, 1980; Parsons, 1985).
Thus, many researchers have come to the view that in order to achieve high levels of productivity, quality, safety, and worker satisfaction and motivation, human and machine skills and intelligence must be integrated. There is a need to complement, not replace, human skills and abilities (Havn, 1990.) New technologies should be skill-enhancing rather than skill-replacing (Kidd, 1990).
Although there has been great investment in advanced automation and robotics in U.S. industry, it has not resulted in a dramatic U.S. lead in manufacturing technology. One reason seems to be that in the United States automation has largely been seen as a chance to reduce the labor force; another reason is the belief that automated equipment is sufficiently intelligent in itself to be run by relatively unqualified, and hence cheap, labor.
By contrast, in both Europe and Japan, there has been a tendency to integrate highly qualified labor (e.g., even graduate engineers) into the control of automated manufacturing equipment.
Recently, both industrialists and academics (Brodner, 1986; Kellso, 1989; Kuo and Hsu, 1990; De Greene, 1991) have claimed that the attempt to substitute machine for human intelligence in industry has failed, and they have called for a better understanding and implementation of a human-machine symbiosis in manufacturing. In a classical paper, Bainbridge (1983) pointed out the ''ironies of automation": only those processes that are well understood can be automated; those processes that are poorly understood are then left to humans, so that progressively the humans face more and more difficult tasks when systems falter or fail.
It is now being realized that advanced automation places great demands on the human workforce and that automation should not amount to "designing the human out of the system" (Shaiken, 1984). The question is, rather, how the special characteristics of humans can be integrated into technological systems so that productivity is optimized from the point of view of production level, quality, and safety (Wall et al., 1987).
Sanderson (1989) pointed out a number of aspects of discrete manufacturing in which humans play a central role that is only imperfectly understood, even at the level of the individual operator. To support the integration of humans into manufacturing systems requires that designers, trainers, procedure writers, and all levels of management understand the nature of the skills of the human operator at many levels. This means understanding traditional problems of interface design at the lowest level, through complex activities such as planning, scheduling, expediting, and maintenance, up to the most global level of policy setting by management. Indeed, the concept of macro-ergonomics has recently been introduced. This refers to the extremely important role played by managerial and organizational factors in the efficiency of automated and hybrid systems, and to the dynamics of groups, teams, and crews. It provides a top-down ergonomics to complement the traditional bottom-up ergonomics of design.
The European Economic Community (EEC) has undertaken a massive investment in the application of human factors to industrial and manufacturing systems of the future. This can be expected to give the Europeans a large lead in integrating human intelligence into manufacturing production, something that will have a great impact on their global competitiveness. In the United States there are few signs of a change in thinking; many still see automation as nothing more than a way to employ inexpensive and relatively unskilled labor to run sophisticated machines. But all the indicators are that this will lead to a disastrous failure in competitive manufacturing.
There are two main themes that arise from the considerations discussed above. The first is a drive for research that will improve productivity and
efficiency in industry. The second is a need for research toward a work climate, culture, and environment that will be humane and fulfilling to the workforce. Thus, all levels of ergonomics will have to be studied. These include the interface design for flexible manufacturing systems and computer-integrated manufacturing, as well as the role of humans in planning and scheduling, in design, in maintenance, and in supervisory control as applied to discrete manufacturing systems. In addition, we need models of humans in discrete manufacturing situations, and a consideration of macroergonomic factors, both theoretical and applied, to support design and systems development.
Skill Requirements in Advanced Manufacturing
As automation advances, the workforce required to operate and maintain it splits into two job classes. One class requires a few highly skilled workers. Tasks for these workers are often intellectually challenging. These workers perform the high-technology jobs called for in, for example, advanced flexible manufacturing technology. The other job class requires more workers but places few demands on them for specialized training and education. Their jobs are generally unrewarding, either intellectually or financially. Such jobs may remain in the overall job inventory even in the wake of complete automation.
In the present state of the art, one problem is that, when systems falter or malfunction, humans are called on to perform tasks that may be beyond their capabilities. In such cases, a boring job may suddenly become extremely demanding. It is not clear what level or type of education will best serve manufacturing industry so as to match the task to the human. If the sophistication of advanced machinery calls for highly qualified operators, how are the jobs to be designed so that such people will not be bored and dissatisfied during normal operation? What steps are needed to enable people with a suitable level of education to be satisfied in operating a highly automated plant?
Although relatively few manufacturing jobs can give rise to severe hazards (in contrast to chemical industries, the nuclear industry, etc.), there are certainly problems of both safety and economics when systems fail. It has been said that plants may be designed on the assumption that they will be down for maintenance and reprogramming for as much as 30 percent of their operating time, a situation that is clearly undesirable. We need to know how to maintain operator skills so that time lost due to accidents and system faults is minimized. Supervisory control can cause skills to be lost and leave operators unable to cope with abnormal conditions. Our understanding of what makes operators intervene to take charge of a faltering automated system is rudimentary (Lee and Moray, 1992), and the dynamics
of trust between humans and their machines is far from adequate to define allocation of function or to design systems that optimize their interaction (Zuboff, 1988).
Hirschhorn (1984), Adler (1986), and others have argued that even low-level jobs involving computers require greater responsibility on the part of employees. Errors ramify rapidly through interconnected databases and information systems, and such errors are often hard to reverse. Consequently, care and high-quality work, with attention to detail, are especially important in these jobs. This again raises the issue of sources of error and error prevention in computer-related jobs. We need more empirical studies of performance errors in a range of computer-based jobs in order to design job procedures and whole job systems that prevent or minimize error.
In transaction processing, continuous-flow manufacturing, and several other jobs using information technologies, computers are programmed to take care of routine cases, leaving human operators with a greater mix of trouble-shooting and handling exceptions—employees become monitors, maintenance people, and troubleshooters, rather than "doers." This kind of monitoring work can require different skills: more recognition of patterns, more logical or abstract decision making, more learning from rare events rather than via frequent repetition (Clark et al., 1988:Chapter 4).
Thus, in addition to ergonomic concerns, human factors specialists who seek to understand the demands of these jobs as an input into work design must find new methods of measuring the cognitive workload and cognitive skills of the jobs. Some aspects of cognitive workload and skills are relatively well researched: information overload and competencies in absorbing numerical data and graphical representations. Others are less well understood. For example, many people working with information systems need to visualize the logical structure of the system in order to understand the implications of their own actions, their effects on others, and how to diagnose and correct mistakes. Researchers are at an early stage in understanding system visualization as a skill and as an aspect of cognitive workload (see Carroll and Olson, 1987).
As blue-collar work has become automated, fewer jobs are "hands on"; more and more involve the employee in monitoring production via dials, gauges, and so forth. As Zuboff (1988) has explained, this often leads to a loss of data received directly through the physical senses, resulting in shifts in skills and additional cognitive workload. Machinists, for example, used to depend heavily on tactile skills and sense of vibration when feeding and cutting on traditional machine tools. They cannot use the same senses now that programmable machine tools ("machining stations") are automated and are surrounded by heavy metal hoods. However, some machinists have developed hearing skills to sense when a cutting or milling procedure is going wrong. Even though their machines are muffled by hoods, they seem able to identify slight changes in tone in what seems to outsiders to be a
very noisy machine-shop environment. These clues enable them to stop jobs that are about to go wrong (Attewell, 1992; Zicklin, 1987).
This example demonstrates that skill changes resulting from new technologies can be quite subtle. A simple task analysis would undoubtedly note that the automated machinist needs new programming skills, but might well miss the tactile-to-aural shift. Researchers have hardly begun to describe and make an inventory of these higher-level cognitive skills, let alone begun to design job systems with them in mind.
The widespread introduction of information technologies and automation introduces a number of significant problems that affect employment, the nature of work, the configuration of the workplace, the performance of workers, and the well-being of the workforce.
Although it was once believed that automation would increase unemployment in the United States, this has generally not been the case. Some traditional jobs—particularly those with low skill and knowledge demands—have declined as functions have been automated; however, new jobs have been created. This has raised a number of areas that human factors researchers have the knowledge to contribute to. Examples include:
making technology more human-centered;
designing systems to make proper use of unique human capabilities and designing aids to support task performance;
making work meaningful (e.g., acceptable workload levels; reducing boredom, fatigue, and job stress; giving a sense of control over the work process); and
attempting to satisfy the need for autonomy, education, and training.
Research is needed to address such questions as the following: How can we monitor and enhance worker trust in automation? How can operators be kept in the loop so that they can respond when their skills are needed? How can we monitor and measure human performance to detect symptoms of stress and performance impairment? How do we avoid operator skill obsolescence from the use of rapidly advancing technologies? What are the effects of individual differences in cognitive style on worker performance with automated systems?
We have argued that computer technologies and the shift to an information economy present a considerable challenge for those involved in work design and for those researchers in human factors and related disciplines
who try to provide a scientific basis for work design. Although a number of topics and areas of research have been raised in this chapter, four central themes stand out:
First, more systematic and detailed descriptive research about the skills—especially the higher-level cognitive skills—that are used by today's high-technology workers is an indispensable base for future design of jobs and technologies. The kinds of task and skill inventories used in the past have not adequately captured these kinds of cognitive and organizational skills—leaving designers to "fly blind."
Second, further research is needed on performance and error characteristics of emerging technology jobs. We must find out what causal mechanisms or features underlie observed differences in performance, especially in errors and error rates, so that work systems can be designed to avoid error where possible and so that errors that cannot be avoided can at least be more easily identified and remedied.
Third, research must investigate informal learning processes and skill acquisition among those working with new technology; this is a relatively neglected but important research topic. Insights gathered from such research should be used both to plan formal training procedures and to improve ease of on-the-job learning.
Finally, additional research is required to identify the full range of stressors within emerging-technology workplaces and their consequences for employee performance, morale, and health. Some of these stressors will be ones that ergonomics is well equipped to study. Others will necessitate adding sociopsychological studies to traditional human factors methodologies.
In the long run one might think that the increase in highly intelligent computers, robotics, and microelectronics will continue to displace human operators. Yet in Europe and Japan this does not seem to be the case. In recent years, factories built in the United States by foreign companies have reached high levels of productivity with surprisingly little automation and robotics. And one manufacturer who opened a state-of-the-art car assembly plant stated that only about 30 percent of operations are worth automating. Beyond that, automation is relatively cost-ineffective, and it is better to use human skills appropriately. If that is the case, human factors and management psychology may be as important as automation engineering in ensuring high productivity. But there is little understanding of these factors, and what understanding there is comes largely from foreign experience, which for cultural reasons may not be transplantable to the United States.
The challenge of automation to the human factors community in future years is to learn how to adapt automation to people in ways that are both human-centered and productive. Human-centered technology is being actively pursued in Japan and Europe. Until about five years ago, little of this
approach was practiced in the United States, but a still timely research question is how to produce human-centered designs in the context of American culture.
Adler, P. 1986 New technology, new skills. California Management Review 29(Fall):9-28. 1991 Capitalizing on new manufacturing technologies: current problems and emergent trends in U.S. industry. Pp. 59-88 in National Academy of Engineering/National Research Council, People and Technology in the Workplace. Washington, D.C.: National Academy Press.
Air Force Studies Board 1982 The Effectiveness of the Air Force Nondestructive Inspection Management. Panel on Nondestructive Inspection, Committee on Mechanical Reliability, Assembly of Engineering, National Research Council. Washington, D.C.: National Academy Press.
Attewell, P.A. 1992 Skill and occupational change in U.S. manufacturing. Chapter 3 in P. Adler, ed., Technology and the Future of Work. New York: Oxford University Press. 1994 Information technology and the productivity paradox. Pp. 13-53 in D.H. Harris, ed., Organizational Linkages: Understanding the Productivity Paradox. Panel on Organizational Linkages, Committee on Human Factors, National Research Council. Washington, D.C.: National Academy Press.
Bainbridge, L. 1982 Displays as a source of task load. Ergonomics 25(4):335. 1983 Ironies of automation. Pp. 129-135 in G. Johannsen and J.E. Rijnsdorp, eds., Analysis, Design and Evaluation of Man-Machine Systems: Proceedings of the IFAC/IFIP/IFORS/IEA Conference, Baden-Baden, Federal Republic of Germany, 27-29 September 1982. New York: Pergamon.
Banker, R.D., S.M. Datar, and C.F. Kemmerer 1991 A model to evaluate variables impacting the productivity of software maintenance projects. Management Science 17:1-18.
Boehm, B.W. 1987 Improving software productivity. IEEE Software Society 20:43-57.
Boehm-Davis, D.A., R.E. Curry, E.L. Wiener, and R.L. Harrison 1983 Human factors of flight-deck automation—NASA/industry workshop. Ergonomics 26:953-961.
Brodner, P., ed. 1986 Proceedings of the IFAC Workshop on Skill-Based Automated Manufactoring . Karlsruhe, West Germany: IFAC.
Brodner, P. 1986 Skill-based manufacturing vs "Unmanned Factory": which is superior? International Journal of Industrial Ergonomics 1:145-153.
Brooks, F.P. 1987 No silver bullet: essence and accidents of software engineering. IEEE Computer Society 20:10-18.
Butera, F., and J. Thurman 1990 Automation and Work Design. A study prepared for the International Labour Office. New York: North-Holland.
Carroll, J.M., and J.R. Olson, eds. 1987 Mental Models in Human-Computer Interaction. Committee on Human Factors, National Research Council. Washington, D.C.: National Academy Press.
Clark, J., I. McLoughlin, H. Rose, and R. King 1988 The Process of Technological Change. New York: Cambridge University Press.
Cole, R.E. 1979 Work, Mobility, and Participation: A Comparative Study of American and Japanese Industry. Berkeley: University of California Press
Computer Science and Technology Board 1990 Scaling up: a research agenda for software engineering. National Science Foundation. Communications of the ACM 33:281-293.
Computer Science and Telecommunications Board 1994 Information Technology in the Service Society. National Research Council. Washington, D.C.: National Academy Press.
De Greene, K.B. 1991 Emergent complexity and person-machine systems. International Journal of Man-Machine Studies 35(2):219-234.
DeMarco, T., and T. Lister 1987 Peopleware: Productive Projects and Teams. New York: Dorset House.
Deming, W.E. 1986 Out of the Crisis. Cambridge, Mass.: MIT Press.
Drucker, P. 1993 Post-Capitalist Society. New York: Harper.
Gilbreth, F.B., and L.M. Gilbreth  1973 Applied Motion Study. Easton, Pa.: Hive Publishing.
Hackman, J.R. 1987 The design of work teams. Pp. 315-342 in J.W. Lorsch, ed., Handbook of Organizational Behavior. Englewood Cliffs, N.J.: Prentice-Hall.
Hammer, M., and J. Champy 1993 Reengineering the Corporation: A Manifesto for Business Revolution . New York: Harper and Row.
Havn, E. 1990 Designing for cooperative work. Pp. 35-42 in W. Karwowski and M. Rahimi, eds., Ergonomics of Hybrid Automated Systems II. New York: Elsevier.
Hill, G.W. 1992 Group versus individual performance: are N + 1 heads better than one? Psychological Bulletin 91:517-539.
Hirschhorn, L. 1984 Beyond Mechanization. Cambridge, Mass.: MIT Press.
Jaikumar, R. 1986 Post-industrial manufacturing. Harvard Business Review 63(November/December):69-76.
Kantowitz, S.H., and R.D. Sorkin 1987 Allocation of functions. Pp. 355-369 in G. Salvendy, ed., Handbook of Human Factors. New York: John Wiley and Sons.
Kellso, J.R. 1989 CIM in action: microelectronics, manufacturer charts course towards true systems integration. Industrial Engineering 21:18-22.
Kidd, P.T. 1990 Information technology: design for human involvement or human intervention? Pp. 417-424 in W. Karwowski and M. Rahimi, eds., Ergonomics of Hybrid Automated Systems II. New York: Elsevier.
Kiesler, S., D. Wholey, and K.M. Carley 1994 Coordination as linkage: the case of software development teams. Pp. 214-239 in D. Harris, ed., Organizational Linkages: Understanding the Productivity Paradox. Panel on Organizational Linkages, Committee on Human Factors, National Research Council. Washington, D.C.: National Academy Press.
Knights, D., H. Wilmott, and D. Collinson 1985 Job Redesign: Critical Perspectives on the Labor Process. Brookfield, Vt.: Gower Publishers.
Kornhauser, A. 1965 The Mental Health of the Industrial Worker. New York: Wiley.
Kotteman, J., and W. Remus 1987 Evidence and principles of functional and dysfunctional decision-support systems. International Journal of Management Science 15(2):135-144.
Kotteman, J., F.D. Davis, and W. Remus In Press Computer-assisted decison-making: performance, beliefs and illusion of control. In Organizational Behavior and Human Decision Processes. New York: Academic Press.
Kuo W., and J.P. Hsu 1990 Update: simultaneous engineering design in Japan. Industrial Engineering 22:23-28.
Lee, J.D., and N. Moray 1992 Operators' monitoring patterns and fault recovery in the supervisory control of a semi-automatic process. Pp. 1143-1147 in Proceedings of the Human Factors Society 36th Annual Meeting. Santa Monica, Calif.: Human Factors Society.
McGrath, J.E. 1984 Groups: Interaction and Performance. Englewood Cliffs, N.J.: Prentice-Hall.
McGregor, D. 1960 The Human Side of the Enterprise. New York: McGraw Hill.
Mills, D.Q. 1991 Rebirth of the Corporation. New York: John Wiley.
Moray, N.P., and B.M. Huey, eds. 1988 Human Factors Research and Nuclear Safety. Panel on Human Factors Research Needs in Nuclear Regulatory Research, Committee on Human Factors, National Research Council. Washington, D.C.: National Academy Press.
National Academy of Engineering 1988 The Technological Dimensions of International Competitiveness . Committee on Technology Issues That Impact International Competitiveness. Washington, D.C.: National Academy Press.
National Academy of Engineering/National Research Council 1991 People and Technology in the Workplace. Washington, D.C.: National Academy Press.
National Research Council 1986 Human Resources Practices for Implementing Advanced Manufacturing Technology. Manufacturing Studies Board. Washington, D.C.: National Academy Press. 1990 The Internationalization of U.S. Manufacturing: Causes and Consequences. Manufacturing Studies Board. Washington, D.C.: National Academy Press.
Nobel, D.F. 1984 Forces of Production: A Social History of Industrial Automation . New York: Alfred A. Knopf.
Orlikowsky, W.J. 1988 Information Technology in Post-Industrial Organizations. Unpublished PhD dissertation. Stern School of Management, New York University.
Parsons, H.M. 1985 Automation and the individual: comprehensive and comparative views. Human Factors 27(February):99-111.
Peters, T., and R. Waterman 1983 In Search of Excellence. New York: Warner Books.
President's Commission on Industrial Competitiveness 1985 Global Competition: The New Reality. Washington, D.C.: U.S. Government Printing Office.
Sanderson, P.M. 1989 The human planning and scheduling role in advanced manufacturing systems: an emerging human factors domain. Human Factors 31(6):635-666.
Savage, C. 1990 Fifth Generation Management. Burlington, Mass.: Digital Press.
Shaiken, H. 1984 Work Transformed: Automation and Labor in the Computer Age. New York: Holt Reinhart Winston.
Simpson, J., and W. Wood 1993 Where is the group in social psychology? An historical overview. In S. Worchell, W. Wood, and J. Simpson, eds., Group Process and Productivity. Newbury Park, Calif.: Sage Press.
Taylor, F. 1911 The Principles of Scientific Management. New York: Harper and Row.
Walker, C., and R. Guest 1952 Man on the Assembly Line. Cambridge, Mass.: Harvard University Press.
Wall, T.T., C.W. Clegg, and N.J. Kemp 1987 The Human Side of Advanced Manufacturing. New York: Wiley.
Wiener, E.L. 1985 Beyond the sterile cockpit. Human Factors 27:75-89.
Wiener, E., and R.E. Curry 1980 Automation in the cockpit: some generalizations. In Proceedings of the Annual Meeting of the Human Factors Society, October. Santa Monica, Calif.: Human Factors Society.
Zicklin, G. 1987 Numerical control machining and the issue of deskilling: an empirical view. Work and Occupations 14:452-466.
Zuboff, S. 1988 In the Age of the Smart Machine: The Future of Work and Power . New York: Basic Books.