- Email: [email protected]
Sociotechnical systems in an age of mass customization
J. Eng. Technol. Manage. 18 (2001) 241–252
Sociotechnical systems in an age of mass customization Larry Hirschhorn∗ , Phillip Noble, Thomas Rankin The Center for Applied Research, Inc., Suite 501, 3600 Market Street, Philadelphia, PA 19104, USA
Abstract Sociotechnical systems theory (STS) emerged as a design tool for democratizing work in an age of mass production. How should STS be re-thought for the age of mass customization? Linking mass customization to the task of building a learning organization, this article examines the case of redesign of a chemical pilot plant whose purpose was to test new equipment and methods for producing new compounds. The redesign brought into sharp relief the ways in which the changing role of the operator and the primary task of learning, leads us to reconsider such basic STS concepts as “autonomy”, “variance control”, and the redundancy of function. We present a new set of concepts and argue that they are more responsive to the challenges of designing learning organizations. We also suggest that the concept of “meaning” should replace the idea of autonomy to express the moral meaning of STS. © 2001 Published by Elsevier Science B.V. Keywords: STS; Mass customization; Learning organization; Autonomy
1. Introduction 1.1. The three pillars Sociotechnical systems theory (STS) was a technical, moral and political discipline. Morally, it was based upon the idea that workers are entitled to working conditions that supported their all around competence and their relationships with work mates, politically it was grounded in the movement for industrial democracy, and technically it offered methods for designing work to minimize the errors or “variances”. These three elements sometimes shaped a contradictory practice. When management sponsored an STS design, it had no intention of relinquishing ultimate control of the plant to the workers. When STS practitioners and consultants helped redesign the plants, they often over-focused on variance analysis, ∗ Corresponding author. Tel: +1-215-382-8607; fax: +1-215-382-8604. E-mail address: [email protected] (L. Hirschhorn).
0923-4748/01/$ – see front matter © 2001 Published by Elsevier Science B.V. PII: S 0 9 2 3 - 4 7 4 8 ( 0 1 ) 0 0 0 3 6 - 4
242
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
and neglected the social psychological needs of workers for autonomy, collegiality and all around competence. 1.2. The impact of STS STS’s impact has been mixed. Japanese lean production methods, based in total quality management (TQM) and just-in-time (JIT) and the American-based business process re-engineering (BPR) have had a more pervasive impact on industrial engineering and plant design. Estimates suggests that after 50 years, STS-like innovations have reshaped only about 30% of manufacturing plants in Canada and the US. (Osterman, 1994; Betcherman et al., 1994). In response, STS practitioners have developed methods for streamlining the design process itself, but it is unclear if these new methods can accelerate the diffusion rate of STS innovations (Lytle, 1997). It is interesting to note that executives appear to favor BPR with its “top down–cost reduction” approach to change over STS and its more inclusive (in terms of participants and goals) approach. This despite the growing empirical evidence that STS designs pay-off (Macy, 1993), and complementary evidence that re-engineering’s effects are spotty. Perhaps, as some have long argued, management’s interest in retaining control outweighs its interest in effectiveness. Indeed, responding to this resistance, STS practitioners rarely make reference to the roots of their practice in industrial democracy. 1.3. Mass customization One hypothesis that explains the limited diffusion of STS is that STS as a philosophy and practice is rooted in the economics of mass production, when the division between blue collar and white collar was hard and fast and companies made money by minimizing costs and using labor efficiently. In this setting, executives and owners worried that any technique even loosely associated with industrial democracy would raise costs and erase the boundary between workers and managers. However, automation, the success of lean production, STS and BPR have all changed the conditions of competition and effectiveness in the manufacturing sector. It is increasingly difficult for companies to sustain their competitive advantage by simply keeping labor costs low. Increasingly, labor is a fixed cost, control systems rather than workers keep production variances within bounds, and digital computers on the shopfloor give workers information on quality and cost that only managers once had. Workers, who were once controlled by management, take on a managerial role by controlling the controls (Hirschhorn, 1984). In this setting, profitability is determined increasingly by the flexibility of the plant, that is the ease with which managers and workers can change the plant’s configuration to produce new products. This new mode of competition becomes increasingly important as companies try to make money by producing high value-adding products designed for the use by a few customers. This process has been called “mass customization”. In other words, looking back, it is now evident that STS matured in a culture of mass production, where semi-skilled workers deployed at work stations managed a workflow to produce a standard product. In the age of mass customization, managers gain less from minimizing labor costs and withholding information from the workers. The times may now be more
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
243
propitious for STS as a plant design philosophy, but what is an STS for the age of mass customization? What design principles do we use when machines control variances, and a small team can oversee an entire factory? Pava (1994) explored this question, suggesting that social deliberations, people talking to people, rather than production processes, should become the focal point for design, but his premature death left this vein of work unexploited.
2. The pilot plant 2.1. A learning organization Mainstream management thinking suggests that in the era of mass customization, we need “learning organizations”. The learning organization is a one that can adapt quickly to new customer demands and marketplace changes, but how do you design a learning organization from the plant floor up? What should it look like? Where are the relevant organization boundaries for creating teams that can learn? What hierarchical relationships, if any are useful? What is the appropriate division of labor? We suggest that we look at the “pilot plant” as a model to help us explore these questions. A pilot plant, built to look, feel and run as an exact but smaller version of its “parent” commercial unit, has a distinctive mission — to test new processes of production and to produce new products. Once completed, engineers can apply the lessons learned to reconfigure a large-scale commercial plant. As a tool of investigation, the pilot plant is unambiguously a learning organization. 2.2. The Baytown plant We were alerted to this possibility when Phillip Noble, one of the three authors of this paper, was put in charge of a sociotechnical redesign of the Polymers Pilot Plant at Exxon’s Chemical’s Baytown Polymers Center, Baytown, TX, USA. Thomas Rankin consulted to the redesign effort and Larry Hirschhorn subsequently studied the impact of the redesign on its functioning. The plant began in 1981 with one pilot unit of “semi-works” size — meaning it could provide small-scale commercial product, when needed, in addition to experimentation for larger facilities that understandably could not afford the time and resources to test new processes, configurations and equipment. The pilot plant workforce was composed of anywhere from 50 to 60 operators, engineers and management. Like its commercial counterpart, it had console operators, “outside” operators, shift supervision and an internal quality control laboratory. It operated 7 days a week, 24 h a day, with 12 h shifts of five to eight people, depending on the nature of the work. It produced a variety of pellets that found their way into a range of plastics and plastic products such as garbage bags. The plant was established during a time when the chemical and oil industry was replacing analogue with digital controls. Computers screens replaced pneumatically and electronically fed gauges in control rooms. Whereas, earlier operators depended on engineers to interpret the analogue data, operators could now read out the state of the system directly from the computer screen. This technical configuration increased the accuracy and speed of control decisions and enabled plant personnel to more finely tune any production run. Exxon needed this capability in order to reduce its dependence on the commodity market. As one manager
244
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
noted, “commodity spells cyclical, and to get around cyclicality we must develop more specialty type applications. This means special properties and finely tuned products. If our customer can tune his process to get a value added product then we as the supplier should be able to do this as well”. The computers offered engineers and operators real time information about the state of the plant during any experimental run. 2.3. A disjunction Though the pilot plant was up-to-date technically, there was nonetheless a significant disjunction between the social and technical dimension of the plant. When the plant started, approximately 50% of the employees were contract employees. These employees had an average length of tenure of about 7 months, while full-time Exxon employees had an average tenure of 2 years. In 1986, Exxon decided to reduce its workforce by offering money to workers who left voluntarily. Management of the pilot plant was shocked to discover that the best employees left. Something was amiss with the plant. Hoping to understand and fix this problem, they sponsored an STS redesign of the plant. Rankin consulted to Noble and the design team of the plant. 3. The scan and the diagnosis 3.1. The disjunction explained Phillip Noble led the team through the traditional technical and social scan of the plant, examining both technical variances and social roles. The “scan”, and Hirschhorn’s subsequent look at the history and functioning of the plant, provided a key insight into the disjunction between the social and technical system. “The plant had been organized and managed as if the employee’s primary task was to maintain steady state conditions of production.” Nothing could have been farther from the truth. The primary task of the plant was to conduct experiments, to test new processes as well as new products. “This meant that the plant was never in a steady state.” Instead, engineers tested the limits of processes often to failure in order to understand the chemistry of the process. In steady state systems, workers keep a process within prescribed limits to minimize the down time, but in an experimental system workers must at times not correct a process going beyond its range. They need to learn what happens when they don not. In this context, down time may be a misleading measure. A plant may always be “up”, but plant personnel may be unable to produce a particular new polymer in any significant quantity. What counts instead is the total amount of time it takes a plant personnel to learn how to produce some quantity of polymer, say a 1000 lb, successfully. In other words, what counts is learning time. Consider the story one engineer told: The consoler operator called a maintenance operator to check the calibration of the analyzer. After hearing, it was off by a significant amount, the console operator asked the maintenance operator to calibrate it. When I found out they had changed it, I almost blew my stack. I asked them why, and they said, ’well, the calibration was wrong.’ So I had to explain to them even though the analyzer was producing wrong information before you
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
245
fixed it, at least it was consistently wrong. Now that you have fixed the analyzer, the data does not mean a whole lot to me because you changed my base line parameters of the 48 h run. In other words, the plant’s purpose is to produce information not products, and sometimes a malfunctioning instrument is better than the one functioning correctly. 3.2. The old role system Why did the console operator mistakenly recalibrate the analyzer? The engineer could blame the operator. He should have known better. However, sociotechnical thinking tells us that an operator’s behavior and awareness is shaped by the social system within which he or she works. Consciousness follows structure. The scan revealed that the plant’s social structure, the roles and relationships that comprised it, made it difficult for workers to understand the purpose of their work. As befits a plant organized to maintain an operation in a steady state, the plant was organized by shifts, with supervisors in charge of each. Yet the unit of analysis was the program; a particular experiment that ran across the shifts over days or months. A shift was a venue for executing a program, and in any shift the plant could be executing several programs. The program was the foreground, the shift the background. Workers made mistakes in part because they occupied roles that that were not synchronized with the work of the plant. In this context, it became clear that much of the “real work” did not take place during a production run, but in fact at briefings before a run and debriefings after. As one employee noted: The key message from the redesign was to highlight what is work, what is valued. It became clear that the pre-run, post-run and interpretive work was as important as getting the catalyst into the system. The social analysis of when and who met showed in particular that there were not enough post-run meetings and those that functioned were understaffed. People began to see that the plant was not just a production operations but was also an R&D setting. The workers in particular were insufficiently engaged in the pre- and post-run activities. 3.3. The supervisory role What was the consequence of this mismatch between the work itself and the social system? As our vignette suggests, this mismatch created tensions between engineers and operators. Not surprisingly, the first-line supervisors were often caught in the middle. Interestingly, during the social scan, most operators complained about the supervisors. They were supposed to manage the handoffs between shifts, but as the operators experienced it, they did this poorly. Yet, when the design team suggested that the supervisory role be eliminated, many operators balked. They wanted the supervisors as buffers between them and the engineers. A vicious cycle was operating here. Operators made mistakes because they were excluded from the fundamental work of the plant. Engineers were frustrated by the operators’ mistakes. In turn, supervisors protected operators from the engineers’ expression
246
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
of frustration. Needing the supervisors as buffers, the operators’ estrangement from the core work of the plant was thus reinforced. This increased the chances that they would make mistakes. If the plant’s social systems and technical system were poorly matched, why did management tolerate this disjunction for so long? We propose two hypotheses. First, they were comfortable with an occupational structure that separated operators from engineers and managers. Second, like many managers, engineers, union officials and others, they did not yet understand how to design a learning organization.
4. New design principles 4.1. The new role system When the plant was redesigned, the mismatch between the social and technical system was eliminated. The experimental program, e.g. “learning how to produce polymer X”, rather than the shift became the anchor for the operators’ sense of place in the system. This was accomplished by assigning one operator on a shift to represent a particular program. Operators from different shifts met periodically with an engineer to plan and debrief runs. This new team, representing a particular program, could now control variances that arose from shift changes, and supervisors were less burdened by the task of controlling inter-shift variances. The redesign clarified why the supervisors had been in a hot seat. They did not have the requisite authority and resources to control the inter-shift variances created by several programs that were being conducted simultaneously. This new design posed another question. How do engineers and operators collaborate, what distinguishes their roles? Traditional STS practice suggested that the engineers become resources to be deployed “as needed” to semi-autonomous teams, but in this case the engineer was in charge of conducting a particular program. Operators did not have the background in chemistry necessary to run an experiment alone. In classical STS thinking, workers are self-managing because they know best how to do the jobs they do every day, but in a pilot plant one cannot say that the operators know best how to conduct experiments. To be sure, management sponsored a three-stage training program for all operators. In the first stage, new hires were trained for 6 weeks, 8 h a day in basic college math and organic chemistry. In the second, 80 h stage, they learned basic processes in controlling and monitoring distillers, boilers, columns and pumps. In the third and final stage, they were placed in the plant under supervision. Yet, this training program could not turn operators into junior engineers. 4.2. Multiskilling versus dynamic complementarity How then could operators add value? Consider the following vignette. The operations manager felt that the plant personnel did not understand the dynamics of the plant as an instrument of experimentation. Neither engineers nor operators knew what kinds of runs were particularly difficult, which equipment failed when and why, what processes when pushed to their limit of pressure or temperature created difficulties, etc. To answer some of these questions, he asked an operator to examine the log books used by operators to record
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
247
data. She found them difficult to decipher. With the manager’s assistance she created her own database to analyze down time. “I had to go back 3 years and take information out of the log books”, she noted, “I pulled up printed information, put it into a Lotus file and then put it into graphs. Using these graphs, I was able to analyze down time and see trends that were normally overlooked.” Plant personnel also faced the problem of managing the “loading” of the plant. How many experiments could it accommodate, what combinations of experiments could it accommodate, was it wise to have experiment B follow A, or will the former complicate the latter, how long did it take to reconfigure equipment between experiments, how long did a changeover take place? “The plant was an investigative tool with dynamics in its own right, independent of the chemistry experiment conducted on it. Operators were responsible for understanding and managing this tool.” These conclusion questions suggest that operators could not operate independently of engineers, semi-autonomously so to speak, but neither could they be junior engineers. Instead, we suggest they were in a relationship of dynamic complementarity with engineers. We use this term to mean that the more effective they were in managing changeovers, reconfiguring equipment, and sequencing and executing runs, the more intensively could the engineers use the plant to test new products and processes. In a relationship of dynamic complementarity, the distinctive competence of each group amplifies the distinctive competence of the other. Batters improve their batting skills when they face good pitchers. 4.3. Redundancy of function The concept of dynamic complementarity cuts across the classical STS conception of the division of labor. In its conception of multiskilling and the “redundancy of function over the redundancy of parts” STS envisions an end to the division of labor. Each worker can do all the tasks. This works when the division of labor is an artifact of political arrangements rather than the outcome of the task system itself. This is why, e.g. multiskilling in traditional plants often fails or succeeds in the degree to which craft unions are willing to surrender their jurisdictions over certain tasks. In the pilot plant, it was utopian to imagine that the difference between engineers and operators could be eliminated, or that the two groups could work uncoupled from each other. Their differences were the starting point for their collaboration. 1 1
Consider the reverse of dynamic complementarity, when each group de-skills the other. An engineer wanted to be called whenever a certain pump went out of statistical control. The operators, with a certain amount of glee, would dutifully call him at all hours of the day. The exchange would go something like Hey, Larry, you asleep? Yeah, it is 02.00 a.m. The X pump went out of SPC control limits. Umm, do you know the reason? Yeah, ol’ Barry left it blocked in. Is that the type of information you rocket scientists need? Thanks (hangs up) Laughter fills the control room as the operator proclaims, “that’ll cure him.” In short, prior to the redesign, it was an on-going battle, an adversarial relationship, with neither side ready to acknowledge the skills or abilities of the other to contribute.
248
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
4.4. Freedom and constraint In classical STS, we put great store on the operators’ tacit knowledge and unformalized practice. Indeed, this was one of the reasons that STS practitioners criticized Taylorism. In searching for “the one best way”, industrial engineers could not possibly anticipate all the creative ways operators had developed to maintain the plant in a steady state. Indeed, by imposing a constricted version of the “one best way” industrial engineers and managers drove operator creativity underground where it would emerge as tricks and games that workers played against the management. Operators focused their attention on the goal of “making out”, securing what they felt was fair pay for a fair day’s work and left issues of plant productivity to first-line supervisors. The pilot plant posed a different problem. If the plant’s goal was to produce interpretable data, there was a risk that operators who developed their idiosyncratic ways of working might create results that could not be compared across shifts. The training coordinator for the plant, who was charged with reviewing written procedures for various jobs, found that they were written in terms that were too general. He gave the following example: One of the jobs an operator does is to periodically go out in the field and measure how much catalyst has been fed. To do this, you keep the catalyst vessel under pressure, give the console operator a starting weight and then a later weight. Those who did it incorrectly, would not pressure up the vessel even though the system itself operates when pressured up. People had their own individual short cuts. Short cuts are the elementary expression of informal work practices. Operators figure out short cuts, because they see connections not anticipated by management or engineers, but in this case, the training coordinator worried that short cuts could introduce variations that would distort data, even if the plant’s steady state was sustained. When a plant is charged with producing a steady state flow of product, procedural variation is helpful. It enables workers to respond to the many contingencies posed by control tasks. However, when the plant is charged with producing viable information, undocumented procedural variation is harmful. As the training coordinator noted: “We are a research setting, and the numbers we give are so critical. If we can not agree on and follow a fixed set of procedures, the engineers are ultimately going to be making wrong decisions”. We are led to ask: does the research nature of the plant’s primary task sneak in the “one best way” through the back door? The answer we suggest is more complicated. For though individual operators lose some discretion over how they do their work, the operators as a whole potentially gain the freedom to learn from their experience and change procedures so that the plant is optimized as a research tool. As we have already argued in this work, the development of the plant as a tool of investigation is the province of the operators. What the operators lose at an individual level, they potentially gain at the level of the group. At the time of this study, the training coordinator wanted to use the operators to build their own dictionary or procedures. Indeed, describing how he had analyzed one procedure he noted: Two operators took a camera out with them and spent some time with a number of operators outside. When we showed the tapes on how operators were lining the cooling system to two different training sessions, a few people said that they were doing things
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
249
Table 1 The two worlds The context Mass production Minimize down time Producing product Maintaining a steady state Work takes place during a run
Mass customization Minimize learning time Producing information Executing a research program Work takes place before, during and after a run
The design principles Mass production Limiting variances The shift as the locus of the team Redundancy of function Specialists (e.g. engineers) as resources to the team Informal work practices Freedom at the level of the procedure Managing boundaries
Mass customization Learning from variances The inter-shift research team Dynamic complementarity Specialists as part of the team Formal work practices Constraint at the level of the procedure Making boundaries
wrong. You see, while you have a majority doing it the same way, some operators are unafraid to do something out of the norm. I think we got resolution on this. By word of mouth, it eventually got around to a consensus. You have to put on enough peer pressure. The work of Brown and Duguid (1996) is helpful here. The training coordinator has conceptualized the operators as constituting a “community of practice”. As a community, they need to assess one another’s work and agree on standards of practice. The resulting set of procedures to be sure imposes constraints, but as the engineers and operators learn to use the plant more productively, as they add reactors, or reconfigure equipment in new ways to produce new polymers, there will always be new challenges to these standards. There is never a final “one best way”. Thus, in this conception, while operators lose the freedom to vary practice according to their own instincts and intuitions, they gain the freedom to create the body of procedures they will all follow. We can rephrase this in sociotechnical terms. In STS, the workgroup is defined by its boundaries. When the boundaries encompass a whole task, the workgroup’s primary task is to reduce the variance “exported” across the boundaries. Workers are free to use any procedures that help them minimize variances. The workgroup must manage its boundary. In the new setting, workers collectively determine the body of rules that constrain their work; they create boundaries. Table 1 presents our characterization of the new design principles for an era of “mass customization”.
5. Joint optimization STS theory argues that in designing a plant we cannot subordinate social requirements to technical ones, nor technical requirements to social ones. This is called the principle of “joint optimization”. We have argued that by redesigning the pilot plant, the design team matched the social structure with the technical system, the work itself. However, this is not
250
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
the same as joint optimization. The latter term refers to the optimization of a joint system in which neither part is subordinate to the other. Jobs should be designed to promote autonomy, e.g. not because autonomy make workers more productive, but because without autonomy people’s needs for fulfilling work will not be met. The system is optimized when the job design simultaneously maximizes productivity and autonomy. The question becomes: What features of the social system should be optimized in the kind of learning organization we have described here? The concept of autonomy may be less useful here because, as we saw, operators must be disciplined in their work practices if they are to produce reliable information. In this context, it is interesting to note that through its social scan, the design team discovered that the operators were isolated from the rest of the company, that in the 9 years of the plant’s existence only one operator ever visited a customer. Engineers were similarly isolated. Frequently, they executed experimental runs to produce samples of a new product that marketing folks would show to prospective customers, but as one engineer noted It is in a minority of cases where marketing people really involve me, and the worst case is when the marketer think he knows what the polymer is and simply says ‘make it for me’, and they don not understand the details. This may seem harsh, but some marketing people feel they don not even need good data to design polymers. 5.1. Meaning This finding suggests that operators and engineers did not have opportunities to experience the purposes of their work, i.e. how what they did was linked to the enterprise’s goals. Working in relative isolation they could not easily understand the context for the work, what the work actually meant to the goals of the enterprise. Indeed, one reason that so many good operators quit when offered the opportunity was that senior management had not sufficiently recognized the plant’s salience to the company’s strategy. This is also why the company was content to rely on contract employees. This perspective suggests that the social system did not help the pilot plant’s members achieve meaning. Indeed, meaning and motivation are closely linked. Industrial psychologists worry about how to motivate workers, not realizing that people need to be motivated to complete a task, only when the task itself has no meaning for them. For example, it is a common occurrence that people can do the most mind-numbing jobs, e.g. stuffing envelopes for a political campaign, when they value the ultimate purpose of their work, in this case electing their candidate. To be sure, STS theory posits that good jobs are meaningful, but in practice STS practitioners have emphasized autonomy over meaning, hence the familiar equation of STS work design with the design of autonomous teams. However, as argued above, in an effective learning organization too much autonomy is dysfunctional. Under conditions of mass customization meaning should replace autonomy as the primary design tool. To be sure, meaningful work can lead to more productive workers, but the human need for meaning transcends the requirements for productivity. In the spirit of joint optimization, this cannot be the rationale for supporting meaning. Instead, our need for meaning lies in our capacity for language. Talking takes us beyond the sphere of immediate experience, where we are satisfied to solve the problems of existence and leads us to consider our
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
251
reasons for being. Language introduces the notion of “why?” Moreover, when people ask “why am I doing this work”, they must make reference to some social norm and social group. “I work to support my children”, “I work to sustain my status among my friends”, “I work to be recognized as a great scientist by my peers”, “I work to buy the goods and services, that make me feel part of the community”. Group life supplies meaning in the course of every-day life. The result is what we call culture. Its absence we call alienation. Indeed, in certain times and places, productivity or abundance can actually create alienation and destroy meaning. In a classical study, one anthropologist found that an isolated tribe’s social structure was built around the scarcity of axes. People who had axes had status, thus, axes were linked directly to how people make meaning of their roles. Upon contact with the outside world, the tribe was able to obtain many steel axes, upsetting the entire status system. This led to significant social disorganization and anomie. Abundance destroyed meaning.
6. Discussion This argument suggests that to jointly optimize the social and technical system, we should design jobs that sustain “meaning creation”. As our case suggests, this means situating the plant in the context of the total enterprise, building roles that take people outside the plant and into interactions with all its stakeholders. It means, as many researchers and practitioners have argued, putting operators into the “strategy conversations” that shape and influence what goals the enterprise chooses. It is no longer a question of simply designing roles within the plant, but roles that take people legitimately into the social system surrounding the plant. It may very well be that by giving people the opportunity to extract meaning from their work, we will also increase their productivity. For example, in one instance a marketing representative at Noble’s insistence presented the business case for a run to a pilot plant team before its start date, emphasizing its importance to the bottom line. The run was executed flawlessly, but in the spirit of joint optimization this is a bonus. Our primary task is to design roles that enable people to extract meaning from their work to help them create and sustain a culture of work. Yet meaning is restless. It does not obey the boundaries we impose on our identities. We may design a social system so that every employee is linked, through their roles, to the widest purposes of the enterprise; its strategy, how it makes money, how it adds value, and yet we may still create alienation. People can project questions of meaning far beyond their work roles. They can ask: What is the purpose of this company? What value does it add to the community I live in? How does my work affect the quality of life of my spouse and my children? What meaning does it have for them? It remains to be seen if STS as a practice is positioned to consider these broader questions. When it first emerged as a discipline its moral roots in a worker’s right to competence and its political roots in industrial democracy enabled its practitioners to reach beyond the narrow issue of industrial efficiency, but the era of mass customization has so up-ended the occupational structure — the distinction between working and managing is slipping away — that STS, a creature of the era of mass production, may slip into history.
252
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
References Betcherman, G., McMullen, K., Leckie, N., Caron, C., 1994. The Canadian Workplace in Transition. IRC Press, Queen’s University. Brown, J.S., Duguid, P., 1996. Stolen knowledge. In: McLellan, H. (Ed.), Situated Learning Perspectives. Educational Technology Publications, Englewood Cliffs, NJ. Hirschhorn, L., 1984. Beyond Mechanization. MIT Press, Cambridge, MA. Lytle, W.O., 1997. Options for accelerated design, unpublished. Osterman, P., 1994. How common is workplace transformation and who adopts it? Industrial and Labor Relations Review 47 (2). Pava, C., 1994. Managing New Office Technology: An Organizational Strategy. Free Press, New York.
Sociotechnical systems in an age of mass customization Larry Hirschhorn∗ , Phillip Noble, Thomas Rankin The Center for Applied Research, Inc., Suite 501, 3600 Market Street, Philadelphia, PA 19104, USA
Abstract Sociotechnical systems theory (STS) emerged as a design tool for democratizing work in an age of mass production. How should STS be re-thought for the age of mass customization? Linking mass customization to the task of building a learning organization, this article examines the case of redesign of a chemical pilot plant whose purpose was to test new equipment and methods for producing new compounds. The redesign brought into sharp relief the ways in which the changing role of the operator and the primary task of learning, leads us to reconsider such basic STS concepts as “autonomy”, “variance control”, and the redundancy of function. We present a new set of concepts and argue that they are more responsive to the challenges of designing learning organizations. We also suggest that the concept of “meaning” should replace the idea of autonomy to express the moral meaning of STS. © 2001 Published by Elsevier Science B.V. Keywords: STS; Mass customization; Learning organization; Autonomy
1. Introduction 1.1. The three pillars Sociotechnical systems theory (STS) was a technical, moral and political discipline. Morally, it was based upon the idea that workers are entitled to working conditions that supported their all around competence and their relationships with work mates, politically it was grounded in the movement for industrial democracy, and technically it offered methods for designing work to minimize the errors or “variances”. These three elements sometimes shaped a contradictory practice. When management sponsored an STS design, it had no intention of relinquishing ultimate control of the plant to the workers. When STS practitioners and consultants helped redesign the plants, they often over-focused on variance analysis, ∗ Corresponding author. Tel: +1-215-382-8607; fax: +1-215-382-8604. E-mail address: [email protected] (L. Hirschhorn).
0923-4748/01/$ – see front matter © 2001 Published by Elsevier Science B.V. PII: S 0 9 2 3 - 4 7 4 8 ( 0 1 ) 0 0 0 3 6 - 4
242
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
and neglected the social psychological needs of workers for autonomy, collegiality and all around competence. 1.2. The impact of STS STS’s impact has been mixed. Japanese lean production methods, based in total quality management (TQM) and just-in-time (JIT) and the American-based business process re-engineering (BPR) have had a more pervasive impact on industrial engineering and plant design. Estimates suggests that after 50 years, STS-like innovations have reshaped only about 30% of manufacturing plants in Canada and the US. (Osterman, 1994; Betcherman et al., 1994). In response, STS practitioners have developed methods for streamlining the design process itself, but it is unclear if these new methods can accelerate the diffusion rate of STS innovations (Lytle, 1997). It is interesting to note that executives appear to favor BPR with its “top down–cost reduction” approach to change over STS and its more inclusive (in terms of participants and goals) approach. This despite the growing empirical evidence that STS designs pay-off (Macy, 1993), and complementary evidence that re-engineering’s effects are spotty. Perhaps, as some have long argued, management’s interest in retaining control outweighs its interest in effectiveness. Indeed, responding to this resistance, STS practitioners rarely make reference to the roots of their practice in industrial democracy. 1.3. Mass customization One hypothesis that explains the limited diffusion of STS is that STS as a philosophy and practice is rooted in the economics of mass production, when the division between blue collar and white collar was hard and fast and companies made money by minimizing costs and using labor efficiently. In this setting, executives and owners worried that any technique even loosely associated with industrial democracy would raise costs and erase the boundary between workers and managers. However, automation, the success of lean production, STS and BPR have all changed the conditions of competition and effectiveness in the manufacturing sector. It is increasingly difficult for companies to sustain their competitive advantage by simply keeping labor costs low. Increasingly, labor is a fixed cost, control systems rather than workers keep production variances within bounds, and digital computers on the shopfloor give workers information on quality and cost that only managers once had. Workers, who were once controlled by management, take on a managerial role by controlling the controls (Hirschhorn, 1984). In this setting, profitability is determined increasingly by the flexibility of the plant, that is the ease with which managers and workers can change the plant’s configuration to produce new products. This new mode of competition becomes increasingly important as companies try to make money by producing high value-adding products designed for the use by a few customers. This process has been called “mass customization”. In other words, looking back, it is now evident that STS matured in a culture of mass production, where semi-skilled workers deployed at work stations managed a workflow to produce a standard product. In the age of mass customization, managers gain less from minimizing labor costs and withholding information from the workers. The times may now be more
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
243
propitious for STS as a plant design philosophy, but what is an STS for the age of mass customization? What design principles do we use when machines control variances, and a small team can oversee an entire factory? Pava (1994) explored this question, suggesting that social deliberations, people talking to people, rather than production processes, should become the focal point for design, but his premature death left this vein of work unexploited.
2. The pilot plant 2.1. A learning organization Mainstream management thinking suggests that in the era of mass customization, we need “learning organizations”. The learning organization is a one that can adapt quickly to new customer demands and marketplace changes, but how do you design a learning organization from the plant floor up? What should it look like? Where are the relevant organization boundaries for creating teams that can learn? What hierarchical relationships, if any are useful? What is the appropriate division of labor? We suggest that we look at the “pilot plant” as a model to help us explore these questions. A pilot plant, built to look, feel and run as an exact but smaller version of its “parent” commercial unit, has a distinctive mission — to test new processes of production and to produce new products. Once completed, engineers can apply the lessons learned to reconfigure a large-scale commercial plant. As a tool of investigation, the pilot plant is unambiguously a learning organization. 2.2. The Baytown plant We were alerted to this possibility when Phillip Noble, one of the three authors of this paper, was put in charge of a sociotechnical redesign of the Polymers Pilot Plant at Exxon’s Chemical’s Baytown Polymers Center, Baytown, TX, USA. Thomas Rankin consulted to the redesign effort and Larry Hirschhorn subsequently studied the impact of the redesign on its functioning. The plant began in 1981 with one pilot unit of “semi-works” size — meaning it could provide small-scale commercial product, when needed, in addition to experimentation for larger facilities that understandably could not afford the time and resources to test new processes, configurations and equipment. The pilot plant workforce was composed of anywhere from 50 to 60 operators, engineers and management. Like its commercial counterpart, it had console operators, “outside” operators, shift supervision and an internal quality control laboratory. It operated 7 days a week, 24 h a day, with 12 h shifts of five to eight people, depending on the nature of the work. It produced a variety of pellets that found their way into a range of plastics and plastic products such as garbage bags. The plant was established during a time when the chemical and oil industry was replacing analogue with digital controls. Computers screens replaced pneumatically and electronically fed gauges in control rooms. Whereas, earlier operators depended on engineers to interpret the analogue data, operators could now read out the state of the system directly from the computer screen. This technical configuration increased the accuracy and speed of control decisions and enabled plant personnel to more finely tune any production run. Exxon needed this capability in order to reduce its dependence on the commodity market. As one manager
244
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
noted, “commodity spells cyclical, and to get around cyclicality we must develop more specialty type applications. This means special properties and finely tuned products. If our customer can tune his process to get a value added product then we as the supplier should be able to do this as well”. The computers offered engineers and operators real time information about the state of the plant during any experimental run. 2.3. A disjunction Though the pilot plant was up-to-date technically, there was nonetheless a significant disjunction between the social and technical dimension of the plant. When the plant started, approximately 50% of the employees were contract employees. These employees had an average length of tenure of about 7 months, while full-time Exxon employees had an average tenure of 2 years. In 1986, Exxon decided to reduce its workforce by offering money to workers who left voluntarily. Management of the pilot plant was shocked to discover that the best employees left. Something was amiss with the plant. Hoping to understand and fix this problem, they sponsored an STS redesign of the plant. Rankin consulted to Noble and the design team of the plant. 3. The scan and the diagnosis 3.1. The disjunction explained Phillip Noble led the team through the traditional technical and social scan of the plant, examining both technical variances and social roles. The “scan”, and Hirschhorn’s subsequent look at the history and functioning of the plant, provided a key insight into the disjunction between the social and technical system. “The plant had been organized and managed as if the employee’s primary task was to maintain steady state conditions of production.” Nothing could have been farther from the truth. The primary task of the plant was to conduct experiments, to test new processes as well as new products. “This meant that the plant was never in a steady state.” Instead, engineers tested the limits of processes often to failure in order to understand the chemistry of the process. In steady state systems, workers keep a process within prescribed limits to minimize the down time, but in an experimental system workers must at times not correct a process going beyond its range. They need to learn what happens when they don not. In this context, down time may be a misleading measure. A plant may always be “up”, but plant personnel may be unable to produce a particular new polymer in any significant quantity. What counts instead is the total amount of time it takes a plant personnel to learn how to produce some quantity of polymer, say a 1000 lb, successfully. In other words, what counts is learning time. Consider the story one engineer told: The consoler operator called a maintenance operator to check the calibration of the analyzer. After hearing, it was off by a significant amount, the console operator asked the maintenance operator to calibrate it. When I found out they had changed it, I almost blew my stack. I asked them why, and they said, ’well, the calibration was wrong.’ So I had to explain to them even though the analyzer was producing wrong information before you
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
245
fixed it, at least it was consistently wrong. Now that you have fixed the analyzer, the data does not mean a whole lot to me because you changed my base line parameters of the 48 h run. In other words, the plant’s purpose is to produce information not products, and sometimes a malfunctioning instrument is better than the one functioning correctly. 3.2. The old role system Why did the console operator mistakenly recalibrate the analyzer? The engineer could blame the operator. He should have known better. However, sociotechnical thinking tells us that an operator’s behavior and awareness is shaped by the social system within which he or she works. Consciousness follows structure. The scan revealed that the plant’s social structure, the roles and relationships that comprised it, made it difficult for workers to understand the purpose of their work. As befits a plant organized to maintain an operation in a steady state, the plant was organized by shifts, with supervisors in charge of each. Yet the unit of analysis was the program; a particular experiment that ran across the shifts over days or months. A shift was a venue for executing a program, and in any shift the plant could be executing several programs. The program was the foreground, the shift the background. Workers made mistakes in part because they occupied roles that that were not synchronized with the work of the plant. In this context, it became clear that much of the “real work” did not take place during a production run, but in fact at briefings before a run and debriefings after. As one employee noted: The key message from the redesign was to highlight what is work, what is valued. It became clear that the pre-run, post-run and interpretive work was as important as getting the catalyst into the system. The social analysis of when and who met showed in particular that there were not enough post-run meetings and those that functioned were understaffed. People began to see that the plant was not just a production operations but was also an R&D setting. The workers in particular were insufficiently engaged in the pre- and post-run activities. 3.3. The supervisory role What was the consequence of this mismatch between the work itself and the social system? As our vignette suggests, this mismatch created tensions between engineers and operators. Not surprisingly, the first-line supervisors were often caught in the middle. Interestingly, during the social scan, most operators complained about the supervisors. They were supposed to manage the handoffs between shifts, but as the operators experienced it, they did this poorly. Yet, when the design team suggested that the supervisory role be eliminated, many operators balked. They wanted the supervisors as buffers between them and the engineers. A vicious cycle was operating here. Operators made mistakes because they were excluded from the fundamental work of the plant. Engineers were frustrated by the operators’ mistakes. In turn, supervisors protected operators from the engineers’ expression
246
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
of frustration. Needing the supervisors as buffers, the operators’ estrangement from the core work of the plant was thus reinforced. This increased the chances that they would make mistakes. If the plant’s social systems and technical system were poorly matched, why did management tolerate this disjunction for so long? We propose two hypotheses. First, they were comfortable with an occupational structure that separated operators from engineers and managers. Second, like many managers, engineers, union officials and others, they did not yet understand how to design a learning organization.
4. New design principles 4.1. The new role system When the plant was redesigned, the mismatch between the social and technical system was eliminated. The experimental program, e.g. “learning how to produce polymer X”, rather than the shift became the anchor for the operators’ sense of place in the system. This was accomplished by assigning one operator on a shift to represent a particular program. Operators from different shifts met periodically with an engineer to plan and debrief runs. This new team, representing a particular program, could now control variances that arose from shift changes, and supervisors were less burdened by the task of controlling inter-shift variances. The redesign clarified why the supervisors had been in a hot seat. They did not have the requisite authority and resources to control the inter-shift variances created by several programs that were being conducted simultaneously. This new design posed another question. How do engineers and operators collaborate, what distinguishes their roles? Traditional STS practice suggested that the engineers become resources to be deployed “as needed” to semi-autonomous teams, but in this case the engineer was in charge of conducting a particular program. Operators did not have the background in chemistry necessary to run an experiment alone. In classical STS thinking, workers are self-managing because they know best how to do the jobs they do every day, but in a pilot plant one cannot say that the operators know best how to conduct experiments. To be sure, management sponsored a three-stage training program for all operators. In the first stage, new hires were trained for 6 weeks, 8 h a day in basic college math and organic chemistry. In the second, 80 h stage, they learned basic processes in controlling and monitoring distillers, boilers, columns and pumps. In the third and final stage, they were placed in the plant under supervision. Yet, this training program could not turn operators into junior engineers. 4.2. Multiskilling versus dynamic complementarity How then could operators add value? Consider the following vignette. The operations manager felt that the plant personnel did not understand the dynamics of the plant as an instrument of experimentation. Neither engineers nor operators knew what kinds of runs were particularly difficult, which equipment failed when and why, what processes when pushed to their limit of pressure or temperature created difficulties, etc. To answer some of these questions, he asked an operator to examine the log books used by operators to record
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
247
data. She found them difficult to decipher. With the manager’s assistance she created her own database to analyze down time. “I had to go back 3 years and take information out of the log books”, she noted, “I pulled up printed information, put it into a Lotus file and then put it into graphs. Using these graphs, I was able to analyze down time and see trends that were normally overlooked.” Plant personnel also faced the problem of managing the “loading” of the plant. How many experiments could it accommodate, what combinations of experiments could it accommodate, was it wise to have experiment B follow A, or will the former complicate the latter, how long did it take to reconfigure equipment between experiments, how long did a changeover take place? “The plant was an investigative tool with dynamics in its own right, independent of the chemistry experiment conducted on it. Operators were responsible for understanding and managing this tool.” These conclusion questions suggest that operators could not operate independently of engineers, semi-autonomously so to speak, but neither could they be junior engineers. Instead, we suggest they were in a relationship of dynamic complementarity with engineers. We use this term to mean that the more effective they were in managing changeovers, reconfiguring equipment, and sequencing and executing runs, the more intensively could the engineers use the plant to test new products and processes. In a relationship of dynamic complementarity, the distinctive competence of each group amplifies the distinctive competence of the other. Batters improve their batting skills when they face good pitchers. 4.3. Redundancy of function The concept of dynamic complementarity cuts across the classical STS conception of the division of labor. In its conception of multiskilling and the “redundancy of function over the redundancy of parts” STS envisions an end to the division of labor. Each worker can do all the tasks. This works when the division of labor is an artifact of political arrangements rather than the outcome of the task system itself. This is why, e.g. multiskilling in traditional plants often fails or succeeds in the degree to which craft unions are willing to surrender their jurisdictions over certain tasks. In the pilot plant, it was utopian to imagine that the difference between engineers and operators could be eliminated, or that the two groups could work uncoupled from each other. Their differences were the starting point for their collaboration. 1 1
Consider the reverse of dynamic complementarity, when each group de-skills the other. An engineer wanted to be called whenever a certain pump went out of statistical control. The operators, with a certain amount of glee, would dutifully call him at all hours of the day. The exchange would go something like Hey, Larry, you asleep? Yeah, it is 02.00 a.m. The X pump went out of SPC control limits. Umm, do you know the reason? Yeah, ol’ Barry left it blocked in. Is that the type of information you rocket scientists need? Thanks (hangs up) Laughter fills the control room as the operator proclaims, “that’ll cure him.” In short, prior to the redesign, it was an on-going battle, an adversarial relationship, with neither side ready to acknowledge the skills or abilities of the other to contribute.
248
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
4.4. Freedom and constraint In classical STS, we put great store on the operators’ tacit knowledge and unformalized practice. Indeed, this was one of the reasons that STS practitioners criticized Taylorism. In searching for “the one best way”, industrial engineers could not possibly anticipate all the creative ways operators had developed to maintain the plant in a steady state. Indeed, by imposing a constricted version of the “one best way” industrial engineers and managers drove operator creativity underground where it would emerge as tricks and games that workers played against the management. Operators focused their attention on the goal of “making out”, securing what they felt was fair pay for a fair day’s work and left issues of plant productivity to first-line supervisors. The pilot plant posed a different problem. If the plant’s goal was to produce interpretable data, there was a risk that operators who developed their idiosyncratic ways of working might create results that could not be compared across shifts. The training coordinator for the plant, who was charged with reviewing written procedures for various jobs, found that they were written in terms that were too general. He gave the following example: One of the jobs an operator does is to periodically go out in the field and measure how much catalyst has been fed. To do this, you keep the catalyst vessel under pressure, give the console operator a starting weight and then a later weight. Those who did it incorrectly, would not pressure up the vessel even though the system itself operates when pressured up. People had their own individual short cuts. Short cuts are the elementary expression of informal work practices. Operators figure out short cuts, because they see connections not anticipated by management or engineers, but in this case, the training coordinator worried that short cuts could introduce variations that would distort data, even if the plant’s steady state was sustained. When a plant is charged with producing a steady state flow of product, procedural variation is helpful. It enables workers to respond to the many contingencies posed by control tasks. However, when the plant is charged with producing viable information, undocumented procedural variation is harmful. As the training coordinator noted: “We are a research setting, and the numbers we give are so critical. If we can not agree on and follow a fixed set of procedures, the engineers are ultimately going to be making wrong decisions”. We are led to ask: does the research nature of the plant’s primary task sneak in the “one best way” through the back door? The answer we suggest is more complicated. For though individual operators lose some discretion over how they do their work, the operators as a whole potentially gain the freedom to learn from their experience and change procedures so that the plant is optimized as a research tool. As we have already argued in this work, the development of the plant as a tool of investigation is the province of the operators. What the operators lose at an individual level, they potentially gain at the level of the group. At the time of this study, the training coordinator wanted to use the operators to build their own dictionary or procedures. Indeed, describing how he had analyzed one procedure he noted: Two operators took a camera out with them and spent some time with a number of operators outside. When we showed the tapes on how operators were lining the cooling system to two different training sessions, a few people said that they were doing things
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
249
Table 1 The two worlds The context Mass production Minimize down time Producing product Maintaining a steady state Work takes place during a run
Mass customization Minimize learning time Producing information Executing a research program Work takes place before, during and after a run
The design principles Mass production Limiting variances The shift as the locus of the team Redundancy of function Specialists (e.g. engineers) as resources to the team Informal work practices Freedom at the level of the procedure Managing boundaries
Mass customization Learning from variances The inter-shift research team Dynamic complementarity Specialists as part of the team Formal work practices Constraint at the level of the procedure Making boundaries
wrong. You see, while you have a majority doing it the same way, some operators are unafraid to do something out of the norm. I think we got resolution on this. By word of mouth, it eventually got around to a consensus. You have to put on enough peer pressure. The work of Brown and Duguid (1996) is helpful here. The training coordinator has conceptualized the operators as constituting a “community of practice”. As a community, they need to assess one another’s work and agree on standards of practice. The resulting set of procedures to be sure imposes constraints, but as the engineers and operators learn to use the plant more productively, as they add reactors, or reconfigure equipment in new ways to produce new polymers, there will always be new challenges to these standards. There is never a final “one best way”. Thus, in this conception, while operators lose the freedom to vary practice according to their own instincts and intuitions, they gain the freedom to create the body of procedures they will all follow. We can rephrase this in sociotechnical terms. In STS, the workgroup is defined by its boundaries. When the boundaries encompass a whole task, the workgroup’s primary task is to reduce the variance “exported” across the boundaries. Workers are free to use any procedures that help them minimize variances. The workgroup must manage its boundary. In the new setting, workers collectively determine the body of rules that constrain their work; they create boundaries. Table 1 presents our characterization of the new design principles for an era of “mass customization”.
5. Joint optimization STS theory argues that in designing a plant we cannot subordinate social requirements to technical ones, nor technical requirements to social ones. This is called the principle of “joint optimization”. We have argued that by redesigning the pilot plant, the design team matched the social structure with the technical system, the work itself. However, this is not
250
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
the same as joint optimization. The latter term refers to the optimization of a joint system in which neither part is subordinate to the other. Jobs should be designed to promote autonomy, e.g. not because autonomy make workers more productive, but because without autonomy people’s needs for fulfilling work will not be met. The system is optimized when the job design simultaneously maximizes productivity and autonomy. The question becomes: What features of the social system should be optimized in the kind of learning organization we have described here? The concept of autonomy may be less useful here because, as we saw, operators must be disciplined in their work practices if they are to produce reliable information. In this context, it is interesting to note that through its social scan, the design team discovered that the operators were isolated from the rest of the company, that in the 9 years of the plant’s existence only one operator ever visited a customer. Engineers were similarly isolated. Frequently, they executed experimental runs to produce samples of a new product that marketing folks would show to prospective customers, but as one engineer noted It is in a minority of cases where marketing people really involve me, and the worst case is when the marketer think he knows what the polymer is and simply says ‘make it for me’, and they don not understand the details. This may seem harsh, but some marketing people feel they don not even need good data to design polymers. 5.1. Meaning This finding suggests that operators and engineers did not have opportunities to experience the purposes of their work, i.e. how what they did was linked to the enterprise’s goals. Working in relative isolation they could not easily understand the context for the work, what the work actually meant to the goals of the enterprise. Indeed, one reason that so many good operators quit when offered the opportunity was that senior management had not sufficiently recognized the plant’s salience to the company’s strategy. This is also why the company was content to rely on contract employees. This perspective suggests that the social system did not help the pilot plant’s members achieve meaning. Indeed, meaning and motivation are closely linked. Industrial psychologists worry about how to motivate workers, not realizing that people need to be motivated to complete a task, only when the task itself has no meaning for them. For example, it is a common occurrence that people can do the most mind-numbing jobs, e.g. stuffing envelopes for a political campaign, when they value the ultimate purpose of their work, in this case electing their candidate. To be sure, STS theory posits that good jobs are meaningful, but in practice STS practitioners have emphasized autonomy over meaning, hence the familiar equation of STS work design with the design of autonomous teams. However, as argued above, in an effective learning organization too much autonomy is dysfunctional. Under conditions of mass customization meaning should replace autonomy as the primary design tool. To be sure, meaningful work can lead to more productive workers, but the human need for meaning transcends the requirements for productivity. In the spirit of joint optimization, this cannot be the rationale for supporting meaning. Instead, our need for meaning lies in our capacity for language. Talking takes us beyond the sphere of immediate experience, where we are satisfied to solve the problems of existence and leads us to consider our
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
251
reasons for being. Language introduces the notion of “why?” Moreover, when people ask “why am I doing this work”, they must make reference to some social norm and social group. “I work to support my children”, “I work to sustain my status among my friends”, “I work to be recognized as a great scientist by my peers”, “I work to buy the goods and services, that make me feel part of the community”. Group life supplies meaning in the course of every-day life. The result is what we call culture. Its absence we call alienation. Indeed, in certain times and places, productivity or abundance can actually create alienation and destroy meaning. In a classical study, one anthropologist found that an isolated tribe’s social structure was built around the scarcity of axes. People who had axes had status, thus, axes were linked directly to how people make meaning of their roles. Upon contact with the outside world, the tribe was able to obtain many steel axes, upsetting the entire status system. This led to significant social disorganization and anomie. Abundance destroyed meaning.
6. Discussion This argument suggests that to jointly optimize the social and technical system, we should design jobs that sustain “meaning creation”. As our case suggests, this means situating the plant in the context of the total enterprise, building roles that take people outside the plant and into interactions with all its stakeholders. It means, as many researchers and practitioners have argued, putting operators into the “strategy conversations” that shape and influence what goals the enterprise chooses. It is no longer a question of simply designing roles within the plant, but roles that take people legitimately into the social system surrounding the plant. It may very well be that by giving people the opportunity to extract meaning from their work, we will also increase their productivity. For example, in one instance a marketing representative at Noble’s insistence presented the business case for a run to a pilot plant team before its start date, emphasizing its importance to the bottom line. The run was executed flawlessly, but in the spirit of joint optimization this is a bonus. Our primary task is to design roles that enable people to extract meaning from their work to help them create and sustain a culture of work. Yet meaning is restless. It does not obey the boundaries we impose on our identities. We may design a social system so that every employee is linked, through their roles, to the widest purposes of the enterprise; its strategy, how it makes money, how it adds value, and yet we may still create alienation. People can project questions of meaning far beyond their work roles. They can ask: What is the purpose of this company? What value does it add to the community I live in? How does my work affect the quality of life of my spouse and my children? What meaning does it have for them? It remains to be seen if STS as a practice is positioned to consider these broader questions. When it first emerged as a discipline its moral roots in a worker’s right to competence and its political roots in industrial democracy enabled its practitioners to reach beyond the narrow issue of industrial efficiency, but the era of mass customization has so up-ended the occupational structure — the distinction between working and managing is slipping away — that STS, a creature of the era of mass production, may slip into history.
252
L. Hirschhorn et al. / J. Eng. Technol. Manage. 18 (2001) 241–252
References Betcherman, G., McMullen, K., Leckie, N., Caron, C., 1994. The Canadian Workplace in Transition. IRC Press, Queen’s University. Brown, J.S., Duguid, P., 1996. Stolen knowledge. In: McLellan, H. (Ed.), Situated Learning Perspectives. Educational Technology Publications, Englewood Cliffs, NJ. Hirschhorn, L., 1984. Beyond Mechanization. MIT Press, Cambridge, MA. Lytle, W.O., 1997. Options for accelerated design, unpublished. Osterman, P., 1994. How common is workplace transformation and who adopts it? Industrial and Labor Relations Review 47 (2). Pava, C., 1994. Managing New Office Technology: An Organizational Strategy. Free Press, New York.