Catching Mistakes Earlier
zeptonaut.comThere's an excellent book called Gemba Kaizen that's all about kaizen and the Toyota manufacturing process that inspired a lot of the thoughts in the post. Lots of interesting parallels between manufacturing and software development (e.g. "factory management should lots of time working on the factory floor if they want efficiency wins").
An odd thing about Japan is that despite making great products it has a terrible safety culture. People from the UAW will point out the carnage in the typical Japanese car factory and it still seems little known that Japan was leading the world in nuclear accidents before Fukushima, such as the Tokaimura critical accidents, the time they tried to cover up a fire at the Monju fast reactor, then the time they dropped the refueling machine on top of Monju, just to name a few.
Fukushima itself revealed numerous lapses in safety procedures, including the fact that they built it at that site at all, that they didn’t have adequate emergency generators, that they never tested the isolation condensers (which would have prevented the meltdown and hydrogen explosion at Unit 1) etc. Contrast that to Nine Mile Point here in upstate NY where they test the isolation condensers every chance they get and the operators know exactly how it works.
Many of the Japanese ideas that lead to quality, efficiency and a seemingly harmonious society lead to death in nuclear work. Critical tasks are treated as 無駄 (waste).
And speaking of 無駄, there is a real tension between a heavier process and faster feedback. That missing header takes 30 minutes to fix but if the process to avoid that mistake adds 30 minutes of time you don’t come out ahead. For superficial problems like that the answer is to speed up your cycle and try to get that 30 minutes down to 20.
A better example of a “a stitch in time saves nine” is where people get the data structures behind the application wrong and fixing it is more like 30 weeks.
Definitely have never heard anything about that aspect of Japanese culture - thanks for sharing!
I have experienced the bit of "Japanese working folk being willing (or perhaps expected) to commit their lives fully to their jobs", and I wonder if putting your own health on the line comes as a byproduct of that? Definitely not an aspect of the culture I'd try to encourage.
(1) The UAW is maybe the best union in the world (been on their picket lines) so that makes American auto factories exceptionally safe
(2) As someone whose obsession with anime even annoys people at anime conventions, my take is that irony has a special place in Japanese culture. For instance they say they have filial piety but they have high rates of elder abuse. Allegedly they have a pacifist constitution but they have a large “Japan Self Defense Force”. There is always an outsider and insider view of a situation which is fertile ground for “normalization of deviance”, see
https://en.wikipedia.org/wiki/Uchi–soto
https://en.wikipedia.org/wiki/Honne_and_tatemae
(3) “Continuous improvement” optimizes the happy path at the expense of resilience to exceptional events. For instance just in time production was crippled by the supply chain shocks of the pandemic. At Tokaimura they were mixing nuclear fuel and were (a) making a higher enrichment than they ever did before and (b) had modified their tools to speed production up. (b) is a competitive advantage in most places but in nuclear fuel processing you have to always avoid forming a critical mass and that is done through applying rules to the process. They got away with it… Until they didn’t.
Thanks for sharing your insight in #2: definitely an interesting view that I wouldn't have access to myself!
As for #3, the same thing occurred to me with regards to the supply chain shortages during the pandemic: the highly optimized, just-in-time supply chain that we'd work so hard to build meant that there was almost no room for extraordinary events. When those events do inevitably happen, they're far more disastrous than they would otherwise have been.
I don't think that _all_ continuous improvement needs to be this way: the classic example of "letting the factory floor worker work with a toolsmith to design a better wrench for their job" probably doesn't have negative consequences in extraordinary situations. However, I do think keeping in mind "is this a pure improvement or are we making a tradeoff, and at what cost?" is a worthwhile question to ask.