In the following, the results are presented in a stepwise fashion, zooming in on the details of SteelWorks, Inc. working practices. Beginning with a description of the architectural setting of the facilities, Sect. 4.1 addresses how the physical division of the offices is mirrored in the perceived hierarchical relations between the administrative and production departments. Building on this context, Sect. 4.2 describes the standard communication flow along the hierarchical slope from administrative to production functions; Sect. 4.3 presents how the employees applied workarounds to foster information flows against this hierarchical slope; and Sect. 4.4 further explores how employees used these workarounds to control and effect an information flow from production to planning functions. Finally, Sect. 4.5 portrays how deception was used as an extreme form of workaround to foster an upstream flow of information, and concludes with a discussion of hierarchical inversion and its consequences for systems design.
Oben/unten: A setting of division and hierarchical opposition
The physical set-up of the SteelWorks, Inc. plant is characterised by an architectural division between the production hall and the administrative offices (Figure 2). The boundary between these areas is clearly demarcated by the heavy steel doors to the production hall, which feature signs indicating that security glasses and steel-toed boots are required beyond that point; in contrast, no specialised gear is required for accessing the office area. Thus, the production hall is labelled as a potentially hazardous area, within which employees require protection.
All labour that is directly related to the production of goods is carried out on the ground floor of a large production hall at the rear of the SteelWorks, Inc. facility. The front side of the building, which is attached to the production hall, hosts the administrative offices, spanning three levels. Plant management functions are located on the first level. Here, the operational administrators and the production manager work in a central open-plan office space. The second level hosts the product engineering team. This space is almost exotic, with international and highly academically qualified staff. It is positioned at the most distanced point from the turning lathes in the production hall.
While there are several routes from the office space to the production hall, the latter is most frequently entered via the open-plan office. From the office, with its noise curbing carpet, a hallway with linoleum flooring leads to the heavy steel door with large blue signs indicating the requirement for protective gear. Thus, the linoleum hallway represents a transition space, emphasising the need for a managed boundary between the two organisational areas.
Upon opening the heavy steel door, which requires a strong pull, one must descend a steel staircase to progress into the production hall. From the top of the staircase, one can view the entire machine park, featuring lathe and milling machines and their operators, wearing boiler suits. The atmosphere is strikingly different from the office space: the air is filled with metallic noise, there is a distinct smell of machine oil, and colour-coded lines on the floor mark walkways and parking areas. Thus, entering the production hall requires not only safety gear, but also knowledge of the basic rules of conduct to manoeuvre safely in the space, such as the standard right of way for the forklift truck or the taboo against touching the razor-sharp steel chippings without suitable protection gloves.
The physical divide between the production and administrative functions finds its first verbal representation in the very distinction between ‘the production’ and ‘the administration’. However, the second frequently-used dichotomy to differentiate between these modes is captured in the German terms oben (‘upstairs’ or ‘above’) and unten (‘downstairs’ or ‘below’). The usage of these terms is highly contextual: oben can refer to the open-plan office in general, specific teams or individuals in the open-plan office, the product engineering department, the management team, or simply colleagues occupying a superior position in the hierarchy. More figuratively, oben can be used as a synonym for a lack of awareness and understanding of the reality of the organisation’s grounding work:
At around 8:30 am, my shadowing partner for the day – a production worker – and I stood at the coffee machine in the production hall when the plant manager and his managing team walked in a group across the main walkway of the production hall to start their daily update enquiry round across all teams relevant for production. My shadowing partner followed the small group (which he later labelled as ‘the delegation’) with a gaze, before turning to the coffee selection. While nesting into the coin slot of the coffee machine, he gave me a sideways glance and commented that he wasn’t sure of the usefulness of this activity [the daily update walk], as, in his opinion, ‘they’ were not aware of ‘what is really going on’, by which he meant the real issues faced when handling the turning lathes day in and day out. They simply would have a view ‘from above’ on the processes. (Field notes)
While the above comment, which was delivered with an unmistakably critical undertone, illustrates the organisational divide, it also affirms the existence of a hierarchical structure. The production worker denied that those from ‘above’ were fully aware of what was going on ‘down here’, but at the same time acknowledged the need for functions that looked down on the production processes ‘from above’. This is relevant, as it shows that the production worker understood both positions, oben/unten, as parts of a whole, in the sense of Dumont’s hierarchical opposition. Another relevant issue is foreshadowed by the worker’s statement that the upper management team was not aware of ‘what is really going on’, suggesting that ‘what is really going on’ (i.e. unten) played an important role—and potentially a more important role than oben—in the work processes. This is of particular importance for the practice of workarounds, as discussed in Sect. 4.5.
The physical set-up of the SteelWorks, Inc. building is mirrored in not only the division of administrative versus production functions, but also the hierarchical relationship between oben and unten (Figure 3).
This idea is reinforced by several interactional and structural examples from my research data.
When I casually talked with a member of the open-plan office about my research plan and mentioned that I would be accompanying a colleague at a turning mill the next day, the response was: ‘Are you sure you want to accompany him for a day? I can absolutely not imagine what you would like to do there for a full day – a couple of hours should be enough.’ We bumped into each other again a few days later and he stated that he still couldn’t understand what I might have found interesting enough at the turning mills to observe over a full workday. (Field notes)
As the ‘worthiness’ of my observation was never questioned when I accompanied employees in administrative functions, there was an implication that jobs in the production area were of lesser value. In line with this, the telephone lists displayed on several desks listed only administrative employees by name. In contrast, workers in the production hall had neither an assigned telephone number nor a company email address. On the telephone lists, they were merely listed via function (i.e. ‘incoming goods management’). Finally, to underline this division between ‘the head’ and ‘the hand’ (i.e. mental planning versus manual execution, as depicted by Taylor), the engineering team was referred to as the ‘brain of the company’, as ‘they have the most information about products and everything new comes from here’. At SteelWorks, Inc., this metaphor even applied to the physical structure of the plant, as the engineering team was located in the uppermost area of the building.
One of the production workers related this structural divide to a potentially conflictual consequence for collaboration:
Collaboration? Hmm, from my experience, as soon as there is a door between two sides, there is also a battle. (Interview, production worker)
The above examples suggest that the structural division of the SteelWorks, Inc. building was met by a corresponding verbal representation of a hierarchical relationship between the administrative and production functions (i.e. planning and execution). This organisational divide into the oben/unten dichotomy stimulated a discussion during my feedback presentation (which occurred 6 months after my fieldwork at SteelWorks, Inc.) about other descriptions used for the administrative and production functions. For instance, as the administrative area at Steelworks, Inc. is closer to the street and entrance hall, some of the attendees claimed that they referred to administrative workers as ‘those at the front’ and production hall workers as those ‘at the back’. During my fieldwork, I did not come across these alternative labels in practice, and thus I cannot comment on the context in which they might have been used. Nonetheless, this discussion affirmed that administrative and execution functions were labelled on the basis of the architectural setting, with its clear hierarchical connotations. Clearly, this organisational set-up comprised the relevant context for the communication flows operating within SteelWorks, Inc. and the workarounds associated with these flows.
Brain to hand: Downstream communication and the digital/analogue divide
The results presented in this section illustrate the established information flows concerning boundary management between the administrative and production functions at SteelWorks, Inc. In particular, they demonstrate three relevant findings for the emergence of workarounds: (1) the established communication flow runs downstream along the hierarchical slope, from planning to production functions; (2) the transition point between the administrative and production functions delineates a boundary between digital and analogue information, which renders the manufacturing processes on the shop floor relatively opaque; and (3) the backflow of information from the production hall to the administrative offices is less clearly established, and requires workarounds (Sect. 4.3). As SteelWorks, Inc.’s ERP system does not represent the organisation’s full suite of technology, my initial approach in the fieldwork was to focus more broadly on digital technology. However, the data quickly highlighted that the ERP system, with its process-oriented and functions-encompassing capacity, should be my primary focus.
At SteelWorks, Inc., the technical engineering team is responsible for the construction of all products. For this purpose, they employ digital tools, including computer-aided design (CAD) software. The main outputs of these tools are the technical drawings for each product. These are stored on a central server and linked to the product master data in the ERP system. A key function of the information flow for product manufacturing is order preparation. The order preparation team acts as a link between the construction and production teams, ensuring that ideas from the mechanical engineers are implemented in the manufacturing process. Thus, these employees operate on a long-term planning level, and are responsible for determining the machine with which certain products will be manufactured. On a daily basis, they are responsible for production scheduling—determining which steps of an order should be manufactured, and at which time, at what machine, and (of greatest relevance to this paper) by which worker.
When an order arrives at SteelWorks, Inc., via fax or email, the order preparation department inputs it into the ERP system, from which it can be accessed (after several review steps) by the sales and procurement teams. In parallel, a paper file copy of the order arrives in a physical inbox labelled ‘new orders’, which triggers the order preparation team to incorporate it into the operational planning. Following this, a production order is generated and, if the order is scheduled to be manufactured within the next 3 weeks, a print-out of the order and the corresponding bill of materials is attached to the folder, before it is temporarily stored in a cabinet. Every 2 to 3 days, the paper orders are finalised as work orders for individual components or manufacturing steps, and supplemented with print-outs of any technical drawings required. These orders are manually carried downstairs to the central Kanban board in the production hall. From there, each work order moves along the individual steps in the production process; this means that the paper file is sequentially allocated (by the order preparation team) to individual to-do boards for different machines, with the respective steel objects following along with it.
When the machine operator starts his workday,Footnote 3 he goes to his work order board, where the orders are sorted into plastic trays labelled ‘Prio 1’ to ‘Prio 5’, indicating the sequence of activity. From there, he fetches the file with the work order, takes the print-out with the technical drawing and attaches it with magnets to the iron door of his turning machine. He then mounts the first unmachined part onto the turning mill and starts to programme the required radii, angles, turning speeds, and tools on the control panel of the machine, according to the technical drawing. When the order is finished (depending on the order size and complexity, this might take up to several hours), the workers indicate this on a terminal with a touchscreen, where they log the set-up time required, the actual production time, and the number of parts completed. As soon as they confirm their entries, the information runs back into the ERP system and the workers proceed to the next order.
Although the ERP system plays a managing role in this process, the interplay between digital and analogue modes of communication is particularly striking. The process steps carried out by administrative staff are recorded on both physical paper and in digital ERP records. However, as soon as the orders are prepared for manufacturing, only the paper representations enter the production process.
In the past, we printed all orders as soon as we checked them [in the ERP system]. Now, we only do this for the next 3 weeks. So if there is a change before the orders go downstairs, you can now simply delete it in the system and you don’t have to go down anymore to find the work orders somewhere. (Interview, order preparation team member)
This quotation, together with observations of order preparation team members walking across the shop floor to track down and update or remove the paper copy of a work order, affirms the break in the information flow. As soon as orders are transformed into analogue representations and passed through the steel door, down the stairs, and into the production area, they are detached from and untraceable in the ERP system. This digital/analogue divide occurs at the transition point between the administrative and production teams, and hence the boundary between the planning and execution functions—between oben and unten. The next digital traction point occurs when a worker logs a manufacturing job in a terminal, which turns the information flow back from production to administration.
The downstream communication flow from the verified origin of an order in the ERP system to its executable counterpart in the production hall seems more clearly defined and established than the counter-directional flow. In SteelWorks, Inc., the information flow is primarily designed as unidirectional, running down the hierarchical slope from planning to execution and mirroring the architectural and organisational hierarchy of ‘brain’ over ‘hand’, oben over unten. Furthermore, the data illustrate that the official back-flow of information from the shop floor up the iron staircase is narrowly designed, and dependent on production workers managing the transition from the analogue manufacturing job print-out to a digital representation within an entry terminal. The interdependencies between the work done by the engineers, the order preparation team, and the production workers, and the consequences of the digital/analogue divide, become salient in the workarounds employed to bridge this divide. The following section focuses on the workarounds associated with information management at the interface between planning and production, as the designed back-flow of information via the ERP system does not facilitate cross-functional cooperation.
Workarounds to manage the staircase
During the morning’s cigarette break I stood next to a colleague from the production team. He wanted to know who I was accompanying that day, then asked me: ‘So you’re with the order preparation team today? Do you already know what they are doing?’
I answered: ‘Production planning?’
Production worker: ‘Yes, planning – that’s the right keyword – and they do this far from reality.’ (Field notes)
The above exchange illustrates the disconnect discussed in previous sections, whereby the shop floor was understood as ‘the reality’ and the administrative functions (hyperbolically) as an area of fantasy. Shortly following this conversation, this idea manifested in lived working practice: my shadowing partner from the order preparation team sat in front of his screen and remarked that ‘the reality’ did not always match the ERP system, and a significant proportion of his workday was invested in approximating the system’s numbers to more closely accommodate what he saw as ‘the facts’ from the shop floor.
He spent the morning tracking down a manufacturing job that was originally for 20 pieces, but later increased to 40. However, only 20 pieces were booked into the system and logged in the terminal next to the turning lathe. He discussed the incident with his teammate, who was sure that the 40 pieces had been manufactured, as he remembered that he had directly told the production worker to increase the job to 40 pieces. Looking at the computer screen in the open-plan office, it was impossible for either to draw conclusions about what to enter. So my shadowee went downstairs to the production area and spoke to his colleague at the respective machine, who assured him that he had produced 40 pieces. Back upstairs at his desk, my shadowee entered the quantity ‘40’ into the computer system and commented to me with a sigh: ‘Now I have to trust the statement of the production guy or do a lot of inquiry work.’ When I asked what he meant, he explained that he would have to search for the parts across the entire production hall, but some might have already been taken to assemble new products. Hence, his decision was to overwrite the official booking of 20 pieces to 40 and to close the process. (Field notes)
Throughout the day, my shadowee walked up and down the staircase and subsequently corrected/updated data in the ERP system, seeking alignment between the manufacturing praxis on the shop floor and its digital representation. Instead of relying on the digital information flow from the terminals at the turning lathes into the ERP system, as intended by the process design, the order preparation worker employed several workarounds. As illustrated by the example above, the main workaround was to engage in face-to-face communication with colleagues on the shop floor to collect the relevant information, and to literally carry this information up the stairs to the open-plan office.
A second established workaround to catalyse information flows against the hierarchical slope was to display yellow cards (similar to those used by football referees) to signal and prioritise. The order preparation team displayed yellow cards to trigger an information flow from the production team upwards to the mechanical engineering team. Typically, this occurred when a technical drawing was reported as incorrect or incomplete by a machine operator. In this event, issues with the drawing would be documented on a Post-it note on the printed technical drawing and put on a cabinet in the open-plan office. Once production planners received these print-outs (often several hours later), they would document them digitally via an email response. However, whenever an issue was considered urgent, the yellow card would be put on a carton that was highly visible across the office floor and would be quickly noticed by an assistant in the mechanical engineering department. The assistant would then deliver the file to one of the mechanical engineers on the second floor, for further action (Figure 4).
This practice illustrates that the processing of such information was deemed important enough by the organisation’s employees to establish a workaround process requiring an explicit manifestation (through the yellow card as an artefact) to facilitate an information flow from the manufacturing hand to the manufacturing brain. It is apparent that the designed information flow (at the intersection between the digitally managed planning process and the analogue work orders) ran significantly more smoothly down the staircase than in the opposite direction, as it required several workarounds to flow up the hierarchical slope—literally upstairs towards the administrative functions. The following section sheds further light on situations in which workarounds were employed to control information flows.
Workarounds to maintain opacity
It was early morning, shortly after 6:00 am. The shift had just started. The production hall was already fully lit and in busy operation mode, while most of the administrative offices were gloomy and vacant. I was accompanying a production worker. After milling the first items, he paused for a moment and his gaze turned upwards towards a row of windows at the other end of the hall. They had caught his attention because the lighting there had just turned from a dimmed standby level to full brightness. The windows belonged to the open-plan office area, and my shadowee reacted to the lighting stimulus with a grumbling frown, before continuing to sort the required tools and materials for the next work task. ‘As one can see, the workday over there has also started in this moment’, I commented, in an attempt to grasp the motivation for his reaction.
He stopped again and looked at me: ‘Yes, but that’s about all you can see from here on what’s happening. But the other way round, everyone up there – they can see everything we do here. Like, who’s working what, and you never know who is watching you.’ (Field notes)
The above vignette connects the structure of the SteelWorks, Inc. building to Bentham’s ‘panopticon’—a macro-ergonomic design for prisons and production plants in the 1880s, which allowed for the supervision of many by only a few in control. Michel Foucault later developed Bentham’s design into his concept of ‘panopticism’ (Foucault 1995), which he used as a metaphor for the increasingly individually embodied power relations of modernity, with the constant threat of observation by supervisors. Foucault held that, as subjects under surveillance can never be sure when they are being observed, they adjust their behaviour in accordance with the given norms. Researchers have applied Foucault’s ideas in the context of video surveillance in public spaces (Foth et al. 2014; Rothmann 2017; Veeraraghavan 2013) and HCI researchers have applied them in the context of health tracking devices (Light and Wright 2009), ERP systems (Sia et al. 2002), location-based services in families (Boesen et al. 2010), and negotiations of control and resistance in the workplace (Bain and Taylor 2000).
In the above scene, the production worker comments that the architectural structure of the building allows the employees in the open-plan office to observe the entire production hall without being noticed by the production workers. He voices dissatisfaction with this threat of constant surveillance, in the sense of Foucault’s panopticism. However, his suspicions do not meet with the reality of the lived praxis in the open-plan office, as fewer than a handful of the employees in that office have their desks in positions overlooking the windows to the production hall; others have to move in an unusual way across the floor and enter other teams’ working spaces to achieve this vantage point. In my fieldwork, I only observed such behaviour when employees were looking for a member of their own team in the production hall; not to check on the production workers.
Nonetheless, the production worker’s sentiments are relevant to the workarounds used to manage the shop floor and the data in the ERP system: the production workers’ suspicion and perception of being under permanent surveillance may have bled into a general suspicion of the company’s digital systems, including the production hall terminals used to log the time required for machine set-up and manufacturing. Contrary to Foucault’s argument—that, in a panopticon environment, rules and behavioural norms are internalised and adhered to without explicit control—the SteelWorks, Inc. production workers employed workarounds in their work logging, perhaps in an effort to maintain a protective veil over the shop floor reality.
The practiced workarounds included either logging the total duration of a manufacturing job without indicating a specific duration for the set-up time, or occasionally ‘forgetting’ to log time altogether. Such instances were often contextualised with the comment that the standard set-up time indicated for a job was not representative, as it was not realistic to expect an average or inexperienced worker to perform the task in that timeframe. Production workers also perceived a deep lack of understanding of the nature of their work, as the following quote illustrates:
Some people from the administration seem to think that I simply throw together some raw materials and tools, shake them once, and everything is ready. I always tell them that there is no ‘just quickly’ because everything always takes a while [to manufacture], no matter what. (Field notes)
Connected with these workarounds to protect the actual work situation on the shop floor were workarounds used to shape the sequence of manufacturing jobs. Individual strategies to arrange manufacturing jobs were important to the production workers, who felt undervalued in their expertise with respect to clustering or sequencing jobs in the most efficient way. As they had no influence on the overall planning and scheduling of manufacturing jobs, they employed workarounds by inspecting the paper files for jobs scheduled in the coming days and prioritising or postponing certain jobs, irrespective of the priority folders they had been sorted into by the production preparation team.
We were standing in front of the to-do board and the production worker was selecting the jobs for the day. He pointed at one file and commented: ‘I’ll leave this [job] sitting here until someone shows interest. Before that, I won’t do anything. It’s not suitable for this machine anyway. I would never attempt to manufacture such a job on it.’ (Field notes)
The consequence of such workarounds was similar to that of the workarounds practised by the production planning team in their efforts to connect reality with the ERP system, as described in the previous section. While the workarounds discussed here aimed at maintaining the opacity of the production process, the following section reports an incident in which opacity was used to catalyse an upstream information flow when the material object of the yellow card was considered too weak to facilitate this process.
Deception as a workaround to foster upstream information flows
Late afternoon at the turning lathe: A member of the production planning team approached the production worker I accompanied that day, one of the lathe operators, with a highly urgent manufacturing job that needed to be done immediately, as the client was calling every day to enquire about the order status. After a short discussion, they agreed that the production worker would take care of the job the following morning.
Next morning, 7:30 am: Upon entering the production hall, I met the production planner and casually asked him about the status of the urgent job from the previous day. He took a deep breath and told me that the production worker had started to work on the job that morning, but there was a significant error in the technical drawing and hence he had machined the part incorrectly and it was unusable for further processing. He now had to organise for another raw part to be produced as quickly as possible and for the drawing to be corrected, so they could start anew with the process. […] He needed to organise that immediately, since the customer was putting a lot of pressure on the company. In seemingly high alert, he rushed up the staircase to the open-plan office.
A few minutes later, I approached the production worker and asked him about the urgent job, as I had learned from the production planner that it had to be re-done […] The man took a quick glance around the hall and told me in a low voice with a broad grin on his face that this was not the case. He had noticed the error on the drawing early enough to exclude the critical areas from machining. But, together with his team, they had taken the decision to come up with the story that the part would have to be manufactured anew. He said: ‘If I had turned my head off this morning and just worked according to the drawing, it would have happened like that. Then inevitably a new part would have had to be made.’ He added: ‘We underlings also have to find a way to show that we’re able to think and use our mind down here, too!’
Same day, 9:30 am: The story about the ill-manufactured part was maintained for over 2 hours. But when the increasingly stressed-out production planner checked again with him to enquire whether a raw part with a different alloy might be an option, the production worker finally terminated the drama and dissolved the matter. The production planner registered the news expressionlessly and later commented to me that they had simply played a joke on him. At that moment, the incident was already brought to the prioritised attention of the mechanical engineers and the plant manager. (Field notes)
The event described above did not lead to any negative consequences for the production worker. However, during my next observation in the engineering office and at the plant managers meeting, different options to ensure that the correct technical drawings were attached to manufacturing jobs were discussed. Without this incident, such a discussion would have likely never been raised. Thus, this vignette illustrates how deception could function as an extreme form of workaround to foster the flow of information against the hierarchical slope. By yarning a catastrophe tale based on an incorrect technical drawing for a manufacturing job, the workers instigated sufficient momentum to force their feedback immediately on the mechanical engineering team and, in parallel, the plant manager, situated at the highest hierarchical level. This workaround can be understood as an intensified version of the yellow card workaround, which was similarly used to facilitate an information flow in this direction. The example underlines the perceived relevance of back-loop information flows from the production area to the administrative functions—and at the same time an unmet need in systems and/or process design.
The incident furthermore poignantly shows that the workers were fully aware of their information advantage about the ‘reality’ on the shop floor, and the power granted to them by the opacity they maintained over the production hall. The production worker and his teammates knew that they could successfully employ the catastrophe tale, as details about the actual status of the part’s manufacture were only known to them, and not to members of the administrative functions. At that moment, the production team held significantly more relevant information than their administrative colleagues, and they used this information advantage to trigger a flow of feedback about the insufficient quality of the technical drawings.
It is notable that the production worker was the one who decided to dissolve the situation. For a short moment, he turned the established hierarchy upside-down and highlighted the knowledge advantage of the production workers for manufacturing processes. His remark—‘We’re able to think down here, too!’—vehemently challenged the idea of the hierarchical opposition between the thinking brain and the executing hand, demonstrating that the execution function could operate beyond the task of blindly fulfilling a manufacturing job. The goal of this deceptive workaround was to ensure that feedback would reach recipients at the upper levels in the hierarchy. Hence, the production worker did not question the organisational structure and value system as such, but he merely highlighted the fact that the information expertise lay with the production team during the manufacturing process, and that the dependencies were inversed in this context. This is exactly equivalent to Dumont’s understanding of hierarchical inversion: at certain points in the SteelWorks, Inc. value chain (namely the execution of manufacturing jobs), the dominant hierarchical structure was inversed. In these moments, the production workers held more information power. This highlights that, while a hierarchical set-up may be neither stationary nor static—designed processes and information systems are. Based on these results, I argue for the inclusion of an organisational layer of analysis in any examination of workarounds, takes such circumstances into consideration.