Customer Reviews

Japan's Software Factories: A Challenge to U.S. Management

5 star:		(0)
4 star:		(2)
3 star:		(0)
2 star:		(1)
1 star:		(0)

 

 

This is one of the earliest in a series of books that Cusumano has written on software technology. In this book he cries "Wolf!": the Japanese are so much better at industrial strength software development than Americans, they have a "software factory", etc., etc.

In his subsequent books, especially those on Microsoft and Netscape, Cusumano slowly discovers that the traditional software development process, requirements/specifications/code, etc., e.g. the waterfall model, is *NOT* the model adopted by successful software companies (and, indeed, not the model adopted by many hardware companies). He learns that designs are not something to be churned out by a factory - indeed, if they can be churned out, then they should be reusing exactly the same software.

In some ways the packaged software industry, e.g. Microsoft, supplanted the custom software industry in this timeframe, the time of the PC; Microsoft's process, which Cusumano calls "synchronize and stabilize", may be considered to be JIT (Just In Time) software specification and development. Or, if not Just In Time, As Soon As Possible and No Earlier than Necessary.

While I cannot agree with the conclusions of this book, it is interesting to have on one's bookshelf, to see the evolution of the author's thought over time.

The explanation for what happened to Japanese software factories is in Cusumano's latest book, The Business of Software, which is a more valuable read. The Japanese factories tried to solve the problem of efficiently building custom systems for Japanese customers using mainframes. Cusumano still argues that the factory approach worked well for mainframe software but Japanese programmers didn't have the skills to shift to newer platforms (PCs, workstations). I think the author over-estimated what the Japanese would be able to do. The Japanese are still struggling with old-style development techniques, despite close to zero-bugs, according to recent data from Cusumano. The Indians adopted similar practices (standard dev techniques, reuse, statistical data) but with much better trained people, more adaptable processes, and have been able to handle a wide variety of systems requirements and technologies. I still find Japan's Software Factories a useful look at how Japanese and some U.S. companies made progress in software engineering, particularly their approach to quality control and testing, and reuse. The Indians have gone a step beyond Japan, but they had to start somewhere.

Both of the previous viewers cite Japan's lack of success in the commercial software business as evidence the methodologies here are inherently inferior to either US or Indian approaches. They're not.

In fact, the evidence provided in Cusumano's more recent book, The Business of Software suggests that these methods work very well indeed. In the study reported in that book, Japanese development firms outperform US and Indian firms significantly on productivity and by an enormous margin on quality. I would suggest that this is a direct result of the programs described in this book and the gradual, but continuous improvements they produced.

Japan hasn't fared as well as the US in software products or as well as India in outsourcing, but that has more to do with other factors than with the methods described here. If Japan had India's labor rates, they could compete very well indeed in outsourcing. If they had America's entrepreneurial environment, venture capital community, educational system, language (English) and the world's largest domestic market, they might have done better in software products. As it is, comparing Hitachi's success in the software industry to, say Netscape's, is misleading. Compare them instead to GE, Xerox, HP, etc., none of whom made a dent in the software industry either. Large corporations, almost universally, move too slow to capture an emerging software market.

One final note, these methods are NOT appropriate for a software startup, but as the software industry continues to mature, values such as productivity and quality will become increasingly important and Japan's software factories can deliver those qualities in spades. In that environment, methods geared to fast moving markets may put you at a disadvantage.

I would give this book five stars if it were more current.