A Plague on Both Their Houses

Matthew Garrahan has filed a useful report for the Financial Times on the "sharp increase in the number of safety-related toy recalls by Mattel and other toy manufacturers." The point that Mr. Garrahan wishes to make is that this is not simply a problem of Chinese manufacturing processes. Rather, if one just looks at the numbers, the problem has more to do with flawed design, rather than manufacturing problems:

More than 1m toys were recalled because of concerns about lead paint. However, Mattel has also recalled some 18m toys containing small magnets that could prove harmful if swallowed.

He then reinforces his point by citing an academic study:

Professor Paul Beamish, from the Richard Ivey School of Business, and Hari Bapuji, at the University of Manitoba, analysed 550 toy recalls made in the US since 1988 and found that more than 75 per cent were due to problems that could be attributed to design flaws. Only about 10 per cent related to manufacturing defects.

It goes without saying that, at least when you go "by the book," every stage of a production cycle has its own review process. Manufacturing is generally reviewed by the statistical techniques of quality control, and it is not hard to get up to speed on the problems of statistical sampling and the interpretation of the sample data. The only real questions that should be addressed by Mattel are where the review takes place (at the site of manufacturing is preferable for a variety of reasons) and by whom. Design review, on the other hand, is far less systematic. When I worked for Fuji Xerox at the time when everyone was deep-ending on "knowledge sharing," they were experimenting with an interesting approach to design review in which everyone in the production cycle had the opportunity to contribute, going all the way to service technicians who had to worry about how easy it would be to diagnose problems and replace parts.

Since I no longer work for Fuji Xerox, I do not know whether or not their experiment led to a more general adoption of such an extensive review process. However, a comment from Professor Beamish leads me to believe that Mattel would dismiss such an approach as being too time-consuming:

There are a lot of companies going to market very quickly with toys. It’s highly competitive and sometimes when there is too much emphasis on speed and low cost, things can get sacrificed, whether that’s product safety or product quality.

Once upon a time we were taught that competition was good for the consumer, performing the dual functions of driving quality up and driving prices down. Professor Beamish's observation seems to indicate that, at least in the toy business, this lesson is now a myth (if it was ever true in any business in the first place). Following the trend of objectifying the subject, the customer is nothing more than the source of some data points, fed into some impersonal process driven by an optimization algorithm. This is how it was when Ford was manufacturing its Pinto whose gas tank positioning made for a high risk of explosion on impact. That was forty years ago. These days we have more data points; but the processes of review and decision seem to be just as short-sighted, if not more so.