Toyota Production System as a Cybernetic Management System for Production Processes
Contextual Quote
"The solution that Toyota developed was to institute a sort of ‘social relations bubble’ – to simply protect workers from any negative consequences arising from their self-organized improvements to the production processes, regardless of how costly doing so may appear. They instituted a long standing ‘no layoffs’ guarantee to their production workers, which was observed by the company for many years. I think of this guarantee as the ‘golden rule’ of cooperative relations of production, because it means that work teams can freely improve their processes without concern for any negative implications of doing so. Fundamentally, without this guarantee, it is virtually impossible to cultivate cooperative labor conditions. This guarantee, therefore, is essential to the success of self-organized production, because, in the words of management expert Paul Adler, “90% of people, if you give them a chance to work smarter and improve their jobs, and if they find that by doing that they have created free time for themselves, will spontaneously look for new things to do. I’ve got hundreds of examples [my emphasis].”
This was also Toyota’s conclusion: under the appropriate regime of social relations, people innately enjoy success in challenging situations, especially in the context of group dynamics and teamwork. They found that, using algedonics and the Kanban system, along with the no-layoffs policy, their work teams were extremely innovative at redesigning production processes over and over, in order to keep production going under ever higher standards of precision. This could never have been achieved under traditional means of factory management, which instrumentalizes labor to do a particular and predefined job."
- Dónal Ó Coisdealbha [1]
More at: Conditions for Cooperative Relations of Production
Discussion
DONAL COSTELLO:
"The Toyota Production System (TPS) is about as far as our human industrial civilization has so far advanced in creating a cybernetic management system for production processes. Whereas Stafford Beer demonstrated which functions an optimal cybernetic system needed, the TPS actually built out the functions for their specific economic-technical and social context.
The TPS emerged from the need which Japanese industry faced to catch up with the huge US automobile manufacturers from the 1950s onward. During their tours of the United States, representatives of the Toyota company noticed two important things – firstly that the Fordist and Taylorist manufacturing orthodoxy looked good on the surface but was actually extremely wasteful, and secondly that the key to a superior system lay in how US supermarkets implemented stock control in order to provide the right amount of product to meet customer demand.
It was not a coincidence that this happened in the automobile industry. It was here that the complexity and variety within the production processes reached such a high level, that in order to make significant further improvements, a new management system was objectively required. Toyota discovered, rather than invented, this path forward for high complexity production – the Fordist technique, as we’ll now consider, had already been played out.
* What was the problem with Fordism?
Simply a lack of variety. Fordist production had, from 1908 onward, represented a massive jump in quality and productivity over how factories used to operate, essentially as large workshops. The idea of training people into specialised roles, using standardised tools, to produce exactly the same product over and over on an assembly line, allowed for the first time products of fairly high complexity to be created on a mass scale with a reproducible level of quality.
Workers were regarded here as a type of machine, which could produce a certain number of output parts, given the presence of other machines and a certain number of input parts. This was spelled out explicitly in Frederick Taylor’s Principles of Scientific Management (1911). Production engineers would pre-determine every minute detail of what every worker should do, and then the workers would follow those work steps precisely. That was very successful, and is still a widely used method today, especially in low-technology production processes.
...
most problematic is that emergent problems in the production process do not dynamically force a change in how production works, because there are always plenty of spare components to keep the process going at a high utilisation rate. Even if a part of the process fails for a whole day, this is unlikely to force a change in how the process works, because that would mean a lot of extra, and apparently unnecessary, work. On paper, it doesn’t appear urgent to change anything because if you look at the production data, you’ll see that thanks to inventory buffers, production elsewhere never stopped during a period of unscheduled downtime at any given step in the process.
What I have described above is these days called ‘push’ or ‘batch-and-queue’ production, but is synonymous with the old Fordist system. It is called ‘push’ because each process pushes its output onto the next step as fast as it can, causing a high utilisation rate and low cost-per-unit, but also lots of wasted time, effort and material, and a structurally built-in resistance to change. It is also the ‘natural’ system from the perspective of higher management control, because they assume direct responsibility over each stage in the production process.
* Production from below
The Toyota Production System is fundamentally a self-organising system, and therefore a truly cybernetic organisation of production. In contrast to Taylorism and Fordism, in which production engineers design the minutia of specific production operations which are then carried out to the letter by factory operators, here the operators themselves ‘hold the stopwatch’. Workers on the factory floor are organised into ‘work teams’, with a ‘team leader’ (who is more like a ‘delegate’ than a ‘leader’). The leader’s job is to gain consensus with their team about the best current way to carry through production. The leader is responsible for it, but does not have the authority to keep their new method, if the higher management function of the organisation decides that this method is not appropriate for any reason. The team leader has responsibility, but not ultimate authority.
This means that the production process under the TPS is constantly changing, not according to the direct instructions of the factory’s higher management but rather under their close supervision. The role of the process control engineer now reduces to interfacing with the work team leaders, and standardising the current best practices that they have come up with. On the basis of this process of moving from one standard to the next, while always improving, the TPS enables highly complex production systems to adapt as needed, superseding the rigidity of Fordist production.
It is also important to consider the social-relations aspect of this. The implication of a continually improving production process, from the perspective of the operator, is that they might easily ‘improve themselves out of a job’! This is always a major practical problem with the implementation of the TPS – management think that they can get workers to improve processes, without worrying about the implications of this for their own employment. For the TPS to work properly, the workers must be given guarantees that they will not suffer any negative consequences as a result of their own process improvements. Most famously this is expressed through Toyota’s ‘no layoffs’ policy, which they have at various junctures at least tried to implement.
* Pull, don’t push
The Toyota Production System is a ‘pull’ production system, again, modeled on supermarket stock control. This relies on Deming’s principle that ‘the next step is the customer’. Unlike in a supermarket, here the inventory buffers exist in-between production steps, so that the process takes on the form of a chain of steps, with a buffer separating each ‘supplier’ from its ‘customer’.
Specifically in what is called the ‘Kanban’ system, each buffer is a group of ‘bins’ of components, and each of these bins can contain only a fixed quantity of parts. So the idea is that the ‘supplier’ production step can only take away a bin and refill it, on the condition that the ‘customer’ production step has returned the bin empty. The time available to the supplier to produce the products (called the ‘takt time’) is in this way forced to synchronise with the rate of consumption of the customer."
Internal markets and self-organization as key principles
Dónal Ó Coisdealbha:
"For decades, this problem of insufficient variety in both the Soviet and Western blocs was mostly ignored by industry in spite of the spiraling complexity of production. Eventually it was explicitly recognized and then addressed by Japan’s Toyota Corporation through their discovery and development of ‘pull’ manufacturing. They knew by looking at major producers in the 1950s that this had become a severe problem, leading to poor quality, dysfunction and enormous waste. Big industries of the US, just as in the USSR, were producing large quantities of defective and unwanted components throughout their supply chains. Toyota understood that future production would only become more complex, and that this situation represented a fetter on further development. To break from it they needed to radically increase the level of complexity which they could handle. Indeed, the solution was immediately clear to them. Simply have the workers on the production lines themselves self-organize three closely linked tasks which had become too complicated for any small group of experts to manage: the calculation of production quantities at every step, the scheduling of staffing production lines, and continual re-evaluation and improvement upon how production processes worked. Doing so would allow the industrial engineers and other specialists to think about the bigger picture and only intervene in day-to-day production strategically when a big change is to be introduced, or when some kind of significant problems or opportunities arise.
Such an obvious solution came with an obvious problem however: unless they are explicitly told by some group of experts, how could production line operators possibly know how much of every component to make, and when? The solution was found in borrowing the concept of supermarket replenishment systems, where individual goods are replaced in small quantities whenever they begin to run low on the shop’s shelves. Treating each stage in a factory like a shelf in a supermarket would mean that replenishment is initiated by consumption, rather than being planned in advance according to a schedule. In applying this idea to a factory, US quality pioneer W. Edwards Deming coined the principle which bears his name: “the next process is the customer.” This means that there is no difference conceptually between a product moving between production lines within a firm, moving between firms, or between a retailer and a final consumer. In any of those cases, the preceding process must always supply product at the rate required by the subsequent process. Changes in demand for final consumption goods therefore act to modify the average production levels of all logically prior processes in the supply chain. This immediately and automatically provides individual production lines and whole factories with quantitative production and scheduling information. Whereas production had become too complex for experts to organize centrally, this method allowed each line to regulate its own supply locally in response to changes in demand from the subsequent step, and unlike the former case, it works regardless of the increasing complexity of the processes.
Within today’s modern factories, Toyota’s Kanban system is often used, where a certain number of ‘baskets’ exist as a buffer between production processes. The ‘supplier’ step can fill up these baskets with products, but must then stop when the baskets are full, placing a hard limit on its overproduction. This limit forces the supplier to synchronize its rate of production with the rate of consumption of its customers. If they produce too quickly, the baskets will all be full, halting further production. If they produce too slowly, the customer will come to take products but find that the baskets are all empty, thereby halting further production at the customer’s step. In either case, the production process in general soon breaks down, as prior and subsequent steps all along the supply chain must also halt their production, either due to a lack of supply from the impacted lines or because their own baskets quickly fill up due to lack of demand from the impacted lines. The key point to understand is that these ‘breakdowns’ act as information signals, telling producers when their local productivity is incompatible with the productivity of the supply chain as a whole.
The Toyota Production System (TPS) not only allowed for production to be self-organized by the supply chain, but also satisfied the condition of enabling self-organized process improvement. For example, industrial engineers found a new role in strategically reducing the number of Kanban baskets existing between production steps. With this reduction in the size of the buffer, any mismatch in productivity causes the baskets to fill up or run empty in a proportionally shorter time. Reducing the buffer size does not cause the required rate of supply, in principle, to either rise or fall, but it reduces the tolerance for error, requiring it to be even more closely synchronized with the customers’ rate of demand, to prevent the supply chain from breaking down. These breakdowns resulting from the smaller buffer size expose structural flaws in the production system, a lack of precision which then becomes the basis for a new round of process improvement. The only way to keep production going under the new circumstances may be to move machinery to different parts of the factory, or to change the sequential order of how some parts are assembled. This can produce very interesting results–I myself have seen how, when productivity at some step in the process is too fast for the supply chain in general, the achievement of synchronization can lead production workers to recommend that a big and modern machine be replaced by a smaller and ‘less efficient’ (but more suitable) one. In this way, new investment decisions are based on the information signals that indicate non-compatibility between local and global productivity levels.
The extent to which the flaws in the process are overcome is reflected in the subsequent reduction in the number of breakdowns, and the return to smoothly flowing production at this now higher level of required precision. Again, because of the complexity of modern production, no one is better equipped than the production operators themselves to understand which changes would need to be made in order to keep their own production line operating without breakdowns. In The Toyota Way, Jeffrey K. Liker describes how “Toyota turned scientific management on its head and turned over control of standardisation to work teams.”1 This new approach at Toyota gave rise to the concept of “responsibility without authority,” a phrase coined by Toyota manager John Shook, which applied to work teams on production lines.2 It was these teams, not managers, who were charged with maintaining the stable functioning of their line within the requirements of the supply chain, and were given responsibility for making necessary changes in production processes accordingly. However, they did not have the ultimate authority to maintain those changes if they conflicted with the wider requirements of the factory’s other processes. In this way, local self-organization of the team was reconciled to the global coherence of the system.
The third critical responsibility which was, by necessity, passed over to the workers on the ground was in the area of line scheduling. Under the TPS, labor needs to continually move about, both due to efficiencies being found at different production lines and also due to fluctuations in demand for the products of each production line. As a result, in many modern factories workers are able to sign themselves onto production lines rather than be assigned to a posting by management. Operators will typically have a ‘home line’ at any given time, and can then ‘unlock’ permission to work on other lines by completing hours of training pertaining to those specific lines. Managers, for their part, monitor the extent to which cross-training has given aggregate numbers of workers the appropriate skills to work on each production line, while also tracking the current required number of operators on each of the lines.
This ‘required’ number of operators for any specific line is, in the factory where I work, based on a simple algorithm which incorporates the average number of workers for whom it is the home line, the rate of absenteeism of those workers, the rate of worker turnover on the line and the volatility of demand for the products of that line. For each of the factory’s production lines, a surplus of workers should be seen to exist within the factory who can potentially sign themselves on and work there, over and above the labor which is thought to be strictly required. Given an adequate skills surplus for each line, the production workers are quite capable of self-organizing their schedules with the help of technology: whenever fewer workers than deemed required are currently scheduled to work on a line, this can be seen on the corporate computer software as a flashing warning indicator, alerting all operators who already have the required training to the possibility to sign onto the line and keep production going smoothly."
(https://cosmonautmag.com/2024/08/modern-industry-and-prospects-for-socialism/)
Algedonics
(see also: Stigmergy)
Dónal Ó Coisdealbha:
"A key implication of the self-organization of these three activities – production, process improvement and scheduling – is the profound new importance of what the father of management cybernetics, Stafford Beer, called ‘algedonic signals.’ In a modern context, these signals consist of visual alert icons, again on the corporate computer system, which also generate instant messages by phone and email when activated, and which then require the immediate attention of pre-defined recipients. Such signals are ‘activated’ either manually in light of an immediate problem or automatically, as a result of relevant flows of production statistics crossing over a predefined threshold. These algedonic signals now take on great importance because industrial engineers (as well as other specialists in the fields of quality, safety, and so forth) engage with the pull production system through such exceptional events, when things are going notably well, badly, or when specific classes of issue arise.
In modern factories, when the relevant parties receive such instant ‘algedonic signals,’ they then need to confirm that they will react to them through the corporate computer program, and having done so, they must report on their findings while closing (‘normalizing’) the signal. Both the event and the reaction to it then act as the basis for further discussions and process improvements. Failure by the specialist to react in a timely manner to the problem, or failure to confirm that a problem has been resolved within a given time period, automatically causes the signal to ‘escalate’ to higher levels within the organizational hierarchy, triggering more instant messages. Someone is thus always responsible for any emergent situation, up to and including the top levels of the organization, and inaction as well as any actions constitute information signals which are transparently visible to everyone within the system. The emergent facts of the production process in this way impose accountability on all specialists and managers, subordinating them to the process rather than the other way around."
(https://cosmonautmag.com/2024/08/modern-industry-and-prospects-for-socialism/)