Greetings Economists, I'm slightly confused about the phrase 'industrial'. If we take what Cusemano has written he credits the Japanese agencies to facilitate the software industry creation. He also says there is a natural relationship between creating software, and services developed for the software industry. Microsoft creates software, and IBM provides services might be familiar models. On Jan 30, 2007, at 8:29 AM, Michael Perelman wrote:
Why is Japanese "industrial" software production so much more efficient than large-scale US production? Is it that the Japanese define their goals more narrowly, while the goals of US projects faster than the programmers can program? Is the book about Mitch Kapor's software business worth reading?
Doyle; My impression is the Japanese were flailing around trying to solve the software problem using some modeling from the U.S. industry, but the Cusemano conjecture doesn't really give much insight. A lot of the success with software has been whether or not the hardware supported that software. For example in the U.S. Ajax is conceded to have come along ten years before the hardware performance could justify the protocols. Currently the argument is that software has hit the wall in relation to hardware developments. The problem with bugs or errors in large codes is well known. Microsoft sponsored an internal project in their development of dot net which was meant to address the error prone code problem. This then was abandoned because it competed with dot net. See Intentional Programming. Certain sorts of computing problems have been around for a long time. Parallel processing has been used since the sixties. It is notorious for being hard to 'write' programs for. Intel and others seem ready to put multiple chips in their processors. This means the hardware can address parallel processing in a commodity product. But the solution for writing software for this problem has not been resolved. Therefore the flailing around by the Japanese relates to how well these problems have been conceptualized in society. Let's take parallel processing, in ordinary life we don't think of problems in parallel ways, because our tools of expression do not address those concerns. The computing industry uses for parallel processing have mainly been military (thermo-nuclear modeling of bombs) in the so-called high performance computing, though Google now uses a version of the concept in their server farms. The market for Googles products like finding the right book in all the books of the world is a societal innovation not anticipated by how we use media to figure out what the meaning of life is. Software development seems to me in the U.S. has a relationship to the defense dept. and DARPA that the Japanese don't have. Their high performance computing has generally appeared to be directed more at the civilian market. This may be the key element that distinguished the Japanese industry from the U.S. software community. However, the engineering problem with bad software is not just that sort of social division. thanks, Doyle
