Note: This text was originally the synopsis for a much longer article which I intended to write as the followup to a lightning talk about the subject I did at my workplace. Acknowledging that I probably won’t get time to write the long version, I think this synopsis can stand pretty well on its own as a statement of intent.
DevOps and DevOps-related practices has become a huge thing in the software industry. Elements of this, such as Continuous Integration and Continuous Deployment and the focus on monitoring production systems and metrics has resulted in large improvements in the handling of large-scale deployments. Especially, the act of deployment to production, in traditional systems often an error-prone process riddled with cataclysmic pitfalls and requiring huge amounts of overtime, is reduced to the trivial pushing of a button which can easily be done in normal office hours.
While the success of DevOps largely rests on technological improvements (containerization, orchestration systems, ease of scaling with cloud technologies) as well as process improvements originating in the Agile methodologies as they have developed since 2001 (with concepts such as pair programming, Test Driven Development and a general focus on automatization), much of the literature on DevOps contain a strong “ideological”, to the point of evangelization, promotion of the underlying philosophies of Lean production and management systems. One very conspicuous feature of this ideology is the canonization of Japanese management methods in general and the Toyota Production System (TPS) in particular as an epitome of thoughtful and benign innovation, empowering workers by incorporating their suggestions, achieving world-class production quality while simultaneously showing the maximum respect for each and every one of the humans involved.
This method (the TPS) was, the story goes, introduced in Western manufacturing and later in management, where its basic principles – improvement circles (kaizen), value stream mapping, Kanban, etc. has streamlined the basic business processes, improved productivity and reduced costs. Now, the narrative continues, DevOps will apply these same Lean lessons in the software industry, and we can expect similar vast improvements of productivity.
It is problematic, however, to try to “learn from Toyota” and from Lean Manufacturing without examining in detail how these work in practice, not least how they affect the people actually working in those systems. The authors behind some of the more popular DevOps introductions – The DevOps Handbook and the novels “The Phoenix Project” and “The Unicorn Project” – do not seem to have actually studied the implications of working under the TPS for Toyota’s Japanese employees in great detail, if at all, and seem to have all of their knowledge of the system from American management literature such as James Womack et al’s “The Machine that Changed the World”, basing their own Lean philosophies entirely on Toyota’s own public relations-oriented descriptions of their system.
This is problematic, since it overlooks the distinction between Toyota’s corporate representation of the intention of their production system – and the actual reality felt by automobile workers on the shop floors. Darius Mehri, who worked at Toyota as a computer simulation engineer for three years, has pointed out that the Western management movements inspired by Toyota have failed to understand a very fundamental distinction in Japanese culture and communication: The distinction between tatemae (that which you are supposed to feel or do) and honne (that which you really feel and do). Mehri posits that all Western proponents of The Toyota Way fail to realize that what they are describing is really the tatemae, what management will tell you and what workers will tell you in a formal context when their words might come back and harm them – while the honne is much grittier, much darker and much more cynical.
In effect, proponents of Lean manufacturing and management styles have imported a kind of double-speak in a Japanese variant, but similar to the all too well-known difference between corporate communcations and what workers will confide in private. By doing so, they have inherited the fundamental lie that the priorities of the TPS are respect for each individual employee, partnership between management and workers, and involvement of each and every employee in the continuous improvement of the workplace; while its true priorities are a maximization of profit through the imposition of frenetic work speeds and very long working hours, discarding workers afflicted by the inevitable accidents and work-related diseases – and an “innovation” mainly driven by imitation of other manufacturers.
The truth about the very Toyota Production System that inspired the Lean movement is, leaving the tatemae aside and looking at the honne, that these factories are driven unusually ruthlessly, with little or no regard for the human costs for the workers on the shop floor. Meetings, security briefings and announcements are routinely made after or before actual working hours, when workers are on their own time. Assembly lines are run at extreme speed in order to increase productivity, resulting in serious accidents, chronic work-related diseases as well as production defects. Even so, production targets are set unrealistically high, and the shop crews are not allowed to go home before they are met, often resulting in several hours of daily overtime. The “improvement circles” do exist and workers are indeed asked to contribute, but the end goal is always to increase production and increase line speed, never to create more humane working conditions on the shop floor. Such improvements are (if at all) introduced more grudgingly, e.g. as a consequence of labor shortages and worker dropout.
Lean, by lauding the TPS and uncritically buying its tatemae, is introducing a similar honne of its own: It is, in reality, not revolutionizing productivity, and for all its fair words does not promote the respect of each worker as an individual. On the contrary, the relentless focus on constant “improvements” and constant demand that each employee rationalizes their work as much as possible has caused it to become known as “management by stress”. It may indeed focus on metrics and may indeed choose metrics to demonstrate its own success – while achieving results that range from average/no change to absolutely dismal.
Proponents of DevOps should stop presenting Toyota as any kind of ideal way of working – literally, a nightmarish grind with workers forced to do ten- or eleven hour shifts, ignoring accidents, running beside old and worn-out machinery in outrageously dangerous conditions is not where we want to go. And the “ideal Toyota” with its “improvement kata” and “mutual respect” never existed except as the tatemae to the cynical honne of shop-floor reality. By importing the tatemae as though it were Truth itself, the Lean movement has imported its double-speak – Lean or “management by stress” transitions can be very unpleasant indeed for employees, and while everything is shrouded in talk of partnership and mutual respect, the underlying motivation will often be money-saving through layoffs – the honne to the Lean management bullshit’s tatemae.
That is to say: Perpetuating the lie about Toyota as a humane, innovative and respectful workplace is positively harmful to the employees and processes afflicted by the proposed improvements, as the double-speak involved will inevitably rub off. The Toyota tatemae was not, after all, designed to be practised literally. Accepting it at face value will only set us up for further double-speak in our own practice.
While the software industry can and should continue to evolve based on the philosophy enshrined in the Agile Manifesto and the improved work processes introduced by DevOps, we should eschew the mendacious narrative of Happy Toyota and reject the Lean philosophies that it founded.
Heather Barney and Sheila Nataraj Kirby: Toyota Production System/Lean Manufacturing in “Organizational Improvement and Accountability: Lessons for Education from Other Sectors”, RAND Corporation 2004 (online: https://www.jstor.org/stable/10.7249/mg136wfhf.9).
Ian Hampson: Lean Production and the Toyota Production System – Or, the Case of the Forgotten Production Concepts, Economic and Industrial Democracy & 1999 (SAGE, London, Thousand Oaks and New Delhi), Vol. 20: 369-391 (online: https://library.fes.de/libalt/journals/swetsfulltext/6224179.pdf).
Jeffry S. Babb, Jacob Nørbjerg, David J. Yates, Leslie J. Waguespack: The Empire Strikes Back: The End of Agile as we Know it?, paper given at The 40th Information Systems Research Seminar in Scandinavia: IRIS 2017 – Halden, Norway, August 6-9, 2017 (online: https://research-api.cbs.dk/ws/portalfiles/portal/58521158/IRIS_2017_critical_170501_submission.pdf)
Darius Mehri: The Darker Side of Lean: An Insider’s Perspective on the Realities of the Toyota Production System, Academy of Management Perspectives 20, 2, 2006 (online: https://www.jstor.org/stable/4166230)
Stuart D. Green: The Dark Side of Lean Construction: Exploitation and Ideology, proceedings IGLC-7, 1999, 21-32 (online: https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.22.323&rep=rep1&type=pdf)
Satoshi Kamata: Japan in the Passing Lane: : An Insider’s Account of Life in a Japanese Auto Factory, Pantheon Books, New York (1982)
Gregory A. Howell and Glenn Ballard: Bringing Light to the Dark Side of Lean Construction: A Response to Stuart Green, proceedings IGLC-7, 1999, 33-38 (online: https://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=203907F7926472DB31BBE75D290A826B?doi=10.1.1.418.4301&rep=rep1&type=pdf)
Will Johnson: Lean Production – inside the real war on public education, Jacobin Magazine, December 2012 (online: https://www.jacobinmag.com/2012/09/lean-production-whats-really-hurting-public-education/)
Mike Parker: Management-By-Stress, Catalyst Magazine 1, 2, 2017 (online: https://catalyst-journal.com/2017/11/management-by-stress-parker)
Gene Kim, Jez Humble, Patrick Debois and John Willis: The DevOps Handbook, IT Revolution Press, Portland (OR) 2016.
Gene Kim, Kevin Behr and George Spafford: The Phoenix Project, IT Revolution Press, Portland (OR) 2015.
Gene Kim: The Unicorn Project, IT Revolution Press, Portland (OR) 2019.
Phil Ledbetter: Why Do So Many Lean Efforts Fail?, https://www.industryweek.com/operations/continuous-improvement/article/21144299/why-do-so-many-lean-efforts-fail, 20/9-2020.
Enid Mumford: “Sociotechnical Design: An Unfulfilled Promise or a Future Opportunity”, https://link.springer.com/content/pdf/10.1007/978-0-387-35505-4_3.pdf