There's the old (probably apocryphal?) story of a Japanese firm that got a spec from IBM saying something like "max 3 defects per X units" and in the shipment included a separate, clearly labeled box with 3 defective parts. They didn't understand why, but the customer apparently really cared about those.
Probably not true fully. If they are so detail oriented, they will understand what "max" means and they will also understand how specifications and industry works.
They might have trolled or basically showing off superior production process.
I think they're so used to having things lost in translation, that they take people at their word, no matter how strange it may seem.
Second hand knowledge, but I was told they're detailed oriented because they basically don't trust anyone to tell them when they're wrong. The culture is so adverse to causing offense, that instead of reporting a mistake to a colleague, errors are sometimes allowed through and "engineered around" at the next point of assembly.
There were a lot of nuances to how this sometimes became self correcting and sometimes caused project failure, but as far as I can tell, that applies to everything Japan.
Firsthand sidenote: A similar cultural thing is being ashamed at anything less than proficiency, and will "pretend" to have no knowledge of a topic if they're not an expert. For instance, most people will claim to not understand English, because they're not fluent. But make no mistake, they understand 95% of what you're saying.
There is zero chance they would be able to produce high quality components without understanding what error rates in specifications of other companies mean.
Low error rates in their own output means that they understand industry processes very well.
If the story is true, definitely this. By putting the broken parts in there they pretty much guaranteed that all the other ones are flawless, otherwise they'd have broken the contract.
Maybe they want to test their product with a defective component to see how it behaves and make is more tolerant. When I used to do SSD controller/firmware we used keep out of spec flash memory around to test the robustness of our error recovery. There are ways to simulate the errors but the real stuff always fails in mysterious ways.
The Ford vs. Mazda transmission story on the other hand, is true:
Although the American and Japanese factories were manufacturing the transmissions to the exact same spec, customers were specifically requesting the Japanese transmission. Investigations revealed that while the American transmissions met the spec, the Japanese ones exceeded it by a significant margin.
I have seen things like this. A showroom of defects and mistakes is proof that your quality control system is catching them. It is a statement of professionalism.
Definitely apocryphal, I heard the same story told from a local car manufacturer's perspective (in Austria, so other country and other industry). But it seems back then Japan really made a name for itself w.r.t. quality control. There are more such stories like this going around, like: "Can your process work with such small drills?". The answer: They send back the drill. Drilled.
Except for stories about total engineering fail or borderline fraud, nowadays I couldn't imagine any stories being told about about manufacturing in Germany, US or China.
Speaking of drills, I bought my Makita CX200RB drill set partly because it was manufactured in Japan. I already had a cordless, brushless Makita circular saw, so the alternative would have been something else Makita so I could share batteries. But the fact the units were made in Japan was a big selling point for buying that model.
US seems like a wild mix of rugged and reliable, to shooting themselves in the foot with wildly unsuccessful management and delivering overpriced and underperforming product.
There's still a lot of good (but generally expensive) stuff made in EU and Germany, it's just that "the brand" has been tainted by many companies peddling badly made things with "made in ...".
Fun fact: "Made in Germany" wasn't a sign of quality in the beginning. The British introduced that as a diss, to separate their domestic products from cheap imports.
I can almost see this happening. I used to work at a place that sold hardware (we outsourced production of a custom server unit). Due to the hardware cost and our relatively low volume, we didn't want to spend the money on buying extra units for the developers. Instead, we relied on defective units to do development on.
It would be hard to tune aprocesz to produce exactly 3 defective parts. They would have to target more than 3 per shipment. I guess once they have an inventory of bad parts they can improve their yield until depleted.
As I read it, they copied optional screw fitting holes and badly assembled parts because from an abundance of caution they didn't know what was optional or mandatory.
I watch russion tv and they said that they copied personal photo cameras that pilots took with them. They thought that those cameras may be actually supplied with every US bomber in order to make photos of the enemy territory.
That's just cold war propoganda. They had four B-29s that they cross-referenced in the reverse engineering process, so patches and mis production issues that only affected one were left off.
They did go to great lengths to clone though, but that was a "we can do exactly what you can in two years from essentially nothing" international politics dick measuring contest. Anything other than an exact clone would lead to a "look at how the commies can only make an inferior clone" talking point.
Why would IBM set a hard limit on that, though? Assuming defects are normally distributed, it would make more sense to say: mean of X per N defective parts, with a variance of Y with a P-value of Z.
Though who am I kidding, no one probably evaluates production like that. Though I imagine chip makers are pretty serious about that kind of stuff.
