In September, Liz and I went to Japan for the Ubicomp 2005 conference. It was my first trip to any Asian country, and like any first-time tourist in Japan, I was overwhelmed by the lights, continually surprised by subtle cultural differences (for example, why are so many restaurants in the basements of buildings in Tokyo?) and enchanted by many aspects of the place (the Great Buddha in Kamakura is awesome).
One thing that struck me about Tokyo is how Modernist it is. By Modernist, I mean more than just "new and shiny" or "made of concrete, steel and glass." Much of it seems to be pre- postmodern (in other words, Modernist), in its design. Unironically minimalist housewares and stationary at Muji and Tokyu Hands. 1920s France- and 1960s US-inspired fashion in Harajuku. Sure the cosplay girls dressed like Victorian babydoll zombie dominatrixes is totally not Modernist, but there were few signs of that attitude spilling over into broader life. Even the current fashion in motorcycles seems ultra-minimal. No more splashy plastic and chrome, it's all about having as little as possible between you and the enormous back wheel on your small-engine 1960s bike.
I'm sure I'm missing the point or misunderstanding a lot of what I see, and people who are more familiar with the cultural indicators can point out where I'm misreading the situation, but what struck me is how the Modernist perspective extends to the way that business of design is managed and handled there. Much of the perspective seems based on a classic Modernist "supply-driven" model, which means that a company produces stuff based on internal gut-level determination of what's interesting (usually done by executives or project managers). Sometimes it sells, sometimes it doesn't. When it sells, they make more. When it doesn't, they don't. You can see the Cohen Brothers' version of this Modernist myth of product design is distilled and presented in The Hudsucker Proxy: Norville has a big idea, everyone thinks it's crazy, but they make it anyway, it's a huge hit, he's a genius. No field research, no usability testing, no focus groups. "You know, for kids!" That movie takes place in a fictionalized 30s/50s past that's the glory era of supply-first thinking. In Akihabara, the big electronics district of Tokyo. I got the feeling that's the perspective that Japanese electronics manufacturers still take. Let's make a bunch of variations on an idea; whichever variation sells more, we'll make more like that. It's a continuous Cambrian explosion of products, with the ideas sorted out by the pseudo-evolutionary forces of market adoption.
This is in contrast to the philosophy I see at work at large American and European companies that have enthusiastically adopted end-user research methods taken from marketing techniques and pioneered by the social sciences (see my discussion with Anne recently about the bumps that this adoption is causing in both the academic and corporate spheres). On the one hand, this is great, and the core of my "demand-driven" philosophy of product development. Too many companies are still doing no end-user needs assessment and research at all. Further, I'm certainly not going to say that user research is ever bad ;-). However, the Japanese way of making and marketing made me rethink my stance about the need to always make decisions based on a priori user research. Maybe, just maybe, the current capabilities to prototype, engineer and distribute product variations on a core idea allows for ideas to be tested, and markets to be primed for the acceptance of new ideas, without conclusive documentation from end-user research. Thinking that there's a single product and a single answer, and that research should continue until that's determined, is an equally Modernist idea, from a time when retooling was incredibly expensive. Now, as one hardware designer in San Francisco told me, it's possible to sketch some ideas on a piece of paper, fax it to China, and have a working prototype designed and engineered in a month, and to have production samples soon thereafter. I'm sure this doesn't work for revolutionary ideas, but ideas based on technologies that the engineers and designers are comfortable with--but that's probably where most hardware designs are.
With technological and design possibilities like this, maybe a hybrid approach is appropriate. One that's not based on the idea that the user research is useless ("they don't know what they want" and all that) and also not based on the idea that only deep, exhaustive field research can produce insights that lead to product features. Maybe the hybrid approach is an iterative one based on iteration between rapid research and feature experimentation. What's an appropriate iteration cycle? The Japanese companies have already settled on a release cycle that's nearly quarterly, with a new version of whatever product released every 3 months. However, the feature set in these seems to be still nearly arbitrary, probably determined by product managers in engineering groups based on what the engineers have developed. Maybe the right hybrid approach is a quarterly release cycle of multiple variations on a product idea coupled with both field research to identify new behaviors that can guide innovation and fairly rigorous evaluation of the popularity and actual use of product features. It can be a kind real world conjoint analysis.
Yes, this would screw seriously with the idea that a company needs to put all its marketing muscle behind a single product and promote it as if no others exist. But maybe it's amenable to another approach, one that markets a product line as a single product, but with options. This is the classic automobile model, though one that's fallen out of favor (as I understand it, most Americans buy cars off the lot these days, rather than custom ordering them and waiting, as they did in the old days; Germans, I believe, still prefer to wait and get just the car they want). Maybe a configurator could narrow the options to just the variations that were built and there wouldn't actually be any customization, just assistance in getting the features you want, while follow-up research figures out what you actually use.
Anyway, the thought that struck me as I saw the rows of nearly-identical MP3 players and phones was "There is great research waiting to happen here, if the people making the stuff and the people obsessing about research could only join forces."
Ahh! That answers a lot of questions and explains why a process that seems as chaotic is actually (somewhat) controlled. By designing for themselves, the engineers insure that something is useful to SOMEONE, and not a totally blind guess, but it is definitely brittle outside the core audience of people like them. The reading/not reading of the instructions is also telling, and makes a lot of sense. Thank you for the great, detailed explanation.
The reason Japanese approach works to some degree is the developers really use their products. I mean really use the product, they take it home (if not too large, and being Japanese usually it is not a large product). They won't release it until they have thoroughly satisfied the need. They don't just create something they think will be cool and rush it into production. They create something that was needed and refine it internally. And as you speculate, it does not work well for big innovation. That is why they don't create stuff like the iPod very often. Their process works well to find the hybrid of an already accepted product in most cases.
Your Hudsucker proxy comparison is good, except in Japan it is usually a team effort. One person would not be credited solely with the success. They also had perfected their manufacturing processes decades ago. So it is not really spray-and-pray, but that they more than likely found certain features to be desirable in one instance and not in another and thus worth making because when they do create something successful, it sells big, as you note.
The modifications you see typically don't require massive retrofitting at the factory. Ultimately, even the earlier psuedo-prototypes may sell, especially if there is any shortage of the more popular version. Because the psudo-prototypes are well made, and fit a need. if they don't sell, they didn't invest as much as American companies typically do in developing a new product.
The Japanese consumer is also not adverse to complex products, especially if it will save space or time. So usability takes on a slightly different meaning there. Here is a generalization, that while not always true is maybe also indicative of why the Japanese approach works; US Americans never read instructions, Japanese always read the instructions. Having said that, going to Tokyo and drawing conclusions about what all Japanese are like, is the same as going to only New York or only Los Angeles and drawing conclusions about what US people are like. You will find some difference between the east cost especially Tokyo and the west cost of Japan.
I do like your consept of a hybrid approach, as it provides the benefit of achieving true innovation with the western style of user research and providing greater refinement more quickly using the Japanese way. While somethings in Japan are nearly impossible to change as one poster noted, they quickly adopt successful western patterns/processes and perfect them even further with great enthusiasm.
In another post, Larry Irons makes the following comment:
I can't speak to Japanese consumer culture. However, your suggestion sounds a lot like Don Norman's argument a couple of issues ago in Interactions. He basically argues that breakthrough ideas are more likely to result from the "lone designer" than from customer research. I would simply suggest that the costs of developing a breakthrough product, or let's say ten of them, when only one is needed, or wanted, by the market, reminds me of people who buy penny stocks hoping that one of them will make it big...its more a gamble than a strategy.
To which I reply:
I think that (as I read your description of it) Don Norman's implication is that there's a special kind of insight that happens only in the head of a lone designer. I agree that insight happens inside designers' heads, but I think that innovation is a process that takes those ideas and tests them against the market. Why test against the market, rather than do more research for viability of the ideas? Because there are too many variables to research, and because the market is a dynamic system that changes based on what you put in it (which could support your product idea, or reject it). My theory is that there needs to be a balance between research that narrows and defines an envelope in which to explore, and exploration of that envelope through what Chris calls spray-and-pray product development. It shouldn't be one or the other, but both.
Mike, find me a position where I can do this kind of research and I'm with you all the way - I'm here, in Tokyo, surrounded by these research opportunities :)
Seriously, I think the mood may be starting to change. I fully support the hybrid model, as it seems very possible. However, the change is slow; there's a lot of politics behind this, much of it distinctly Japanese.
This is a great observation, and I hope the hybrid model you're suggesting is where things level out.
The problem with the spray-and-pray approach to cellphones and MP3 players here is that it's hard to develop any sort of loyalty that way. Yes, sometimes some stand-outs emerge, but when they don't, customers are left to make decisions based on minute feature set differences that may not have mattered to them before they left the house.
For smaller, less inspired purchases, a lot of the time I find myself pretty dependent on the knowledge of the sales staff, who are often very good, and very patient at explaining the overly-minute differences.
I'm ok with this because I usually walk out satisfied that I made a logical, rather than emotional purchase, but I'm never thrilled that I got my Sanyo or Hitachi. And worse, since each manufacturer turns out so many mediocre products, I'm no more likely to buy from the same manufacturer again, even if the product performs well.
Muji, who usually only releases one model of every device, is a nice exception, but there is only so much white and beige one can take.