In my statistics class at work we talked about Taguchi Genichi (田口 玄一) recently. Taguchi developed methods for applying statistics to the quality of manufactured goods. This is something the Japanese are very good at, but little appreciated is the fact that is this due to the influence of an American statistician, W. Edwards Deming. Deming tried to introduce better quality control to American industry after World War II, but his work received little notice because American products were in high demand. Basically he was ahead of his time. Industry in the rest of the world had been largely destroyed due to the war, and American workers have always been very productive and capable, whlle the rest of the world had to rebuild and catch up. One of the counties trying to catch up was Japan, and they invited Dr. Deming to teach quality control in Japan. The effects of this are now well-known. Deming refused royalties for his teaching in Japan, so he did not grow rich from this work. Deming's work only became popular in America in the 1980s, after Japan began to overtake America in industrial quality (it took 30 years for this to happen!). However, that is another story.
Taguchi is simultaneously influential and ostracized. Taguchi's philosophy is widely adhered to in American (and worldwide) industry. His methods, strictly considered, are not so popular. Taguchi's philosophy of industrial statistics emphasized the role than uncontrollable factors play in manufacturing. This contrasts with Fisher's experimental designs, which focused on controlling variability through proper design and randomization. Taguchi embraced random effects, because in a manufacturing environment you have less control than in the laboratory. If you don't believe this just ask me sometime. Crazy stuff happens in the industrial world that you just write off as an accident in laboratory work and redo the experiment. In the manufacturing world, you might just have a lot of money invested in that stuff that just turned out weird, so you have to be prepared.
Taguchi emphasizes robustness in your manufacturing process. You do test runs, known as process validations, to encompass as much random variation as you can, because you never know when your supplier will change something, or the humidity is different, or one of the assembly line workers will be having a bad day, beacause you want your process to produce product that meets specification as often as possible. Otherwise you either have to throw it away, or you spend engineering time justifying why the product is acceptable.
Taguchi's methods, however, I cannot claim to understand. From what I know, his experimental designs involve some complicated linear algebra. This has led to controversy in the American statistical community over how to apply these designs. I cannot comment, because I have not looked into them closely. However, if I ever get the chance, I should brush up on my Japanese, because much of Taguchi's work has never really been translated into English. He speaks little English himself, but most international statistics is in English (just like most of science), so there is opportunity here.
One should not be too hard on American manufacturing. The United States produces a great number of high quality products (not that I am biased). For example, while we might buy some of our cars from Japan, Japan buys a lot of military systems from the United States. Regardless of what one thinks of this, the point here is that the product is high quality, higher quality than you can buy anywhere else. In the medical device industry, three quarters of the 30 biggest companies are based in the United States. Not every product is made here, but many, if not most of them are. This is true in a lot of industries. I would personally like to see even more manufacturing in America, but one has to accept that manufacuturing here has higher costs than in Third-World countries. The cost differential is often surprisingly low, given transportation costs, but it is real.