2006/9 Research & Review

The Increasing Complexity of Artifacts and the Role of Product Architecture

TAKIZAWA Hirokazu
Fellow, RIETI

In recent years more than a few academics in the fields of management science and economics have focused their attention on the research theme of what influence product architecture, as typified by modularization, has on industrial organization, corporate strategy, and corporate organization. As a result, a great deal of research has accumulated in this area. Since its establishment, the Research Institute of Economy, Trade and Industry (RIETI) has played a central role in this research in Japan, and even only recently discussion papers by Fujimoto (2005); Oshika and Fujimoto (2006); and Nobeoka, Ito, and Morita (2006), among others, have been posted on the RIETI website.1

This research has shown that Japanese corporate structure excels in its integration capability, and is not really compatible with modularized industries such as IBM PC. Instead, Japanese corporate structure demonstrates its competitiveness in integral-type products like automobiles. On the other hand, however, there is a growing sense of crisis that Japanese companies are losing their international competitiveness in science-based industries, as typified by the semiconductor industry. An example of this is semiconductor lithography. Overall performance in semiconductor lithography is always expected to be greater than the sum of the performance of the individual components, which means that elements requiring integration and tight coordination play a very large part. For this reason, Japanese companies have always been extremely strong in this industry. According to Chuma (2004), however, the international competitiveness of Japanese companies in this field is rapidly declining. Furthermore, Chuma (2006) shows that a decline in international competitiveness of Japanese companies is also occurring in semiconductor production systems.

These phenomena probably have some deep underlying link, but how is it possible to explain them within a unified framework? In order to find the answer to this puzzle, there is perhaps a need to delve even deeper into consideration of the characteristics of innovation in present-day industry from a general perspective. This essay will attempt to approach this topic from the perspective of the increased complexity of artifacts, as well as the division of labor and cooperation between humans and artifacts.

The increased complexity of artifacts and product development

I will start with an explanation of the effect of the increased complexity of artifacts on the present-day development of product systems, based mainly on Okuno, Takizawa, and Watanabe (2006).

Throughout history, human beings have produced and evolved artifacts useful to human activity (Simon, 1996). The constraint regulating the evolution of artifacts at the most basic level is the essential difference between the information processing of the artifacts and the information processing of humans. As will be discussed later, advances in information technology have allowed present-day artifacts to handle information with a precision humans cannot possibly match. Nonetheless, this is purely mechanical, in the sense that this information processing is no more than something carried out according to a prior plan. Humans, on the other hand, are good at contextual information processing that requires the judgment to respond to different situations (Suchman, 1987). This difference represents a gulf between humans and artifacts thus far not bridged by present-day advances in information technology. Given this difference in the information processing of humans and artifacts, we can say humans have manufactured useful artifacts by combining human and artifact information processing in complementary ways.

Although the essential difference in the information processing of humans and artifacts is a gulf that cannot be bridged, IT development in the 20th century have rapidly expanded the range of capabilities of mechanical information processing by artifacts, and have also dramatically cut the costs. As a result of these changes, the process is currently underway of integrating into artifacts the sort of information processing always thought only performable by humans. Present-day artifacts are thus making rapid progress in increasing their own complexity through a pattern of expanding and assimilating the range of mechanical information processing capabilities.

In this regard, there is a need to encapsulate the specialized knowledge used for making artifacts and hide it from the humans that use them, while at the same time it is necessary to make self-explanatory their functionality, operation, and behavior. A further important point is that, due to the bounded rationality of the humans who design and manufacture them, artifacts themselves have a hierarchical structure to their functions, and an approximately matching hierarchical structure to their components; this has meant that artifacts could become increasingly complex.

This has resulted in rapid progress in the hierarchical fragmentation of artifacts, and the appearance of complex product systems comprised of many components. Within the development of these complex product systems, the development of the individual components, as well as the overall operations of coordinating individual components and appropriately integrating product functions, are divided into separate operations and become increasingly specialized. At the same time, products must adequately reflect the demands of the consumer (the artifact user). This means the development of a product system needs to resolve the problem of complex coordination in the three-way standoff between the product developer, component developer, and consumer.

What sort of scheme could resolve the problem of this complex coordination? One possible way is to establish an intangible, "soft" artifact of development standards (i.e., a product architecture), and fix the relationship between the overall product and its components through set-interval commitments in this product architecture to carry out system coordination and system integration. Needless to say, though, this method is not valid for every product system. Coordination through product architecture would be difficult in the case of, for example, automobile manufacture, in which importance is placed on functions such as comfortable ride or design that can only be realized through overall coordination of automotive parts. In such a case, there needs to be coordination by humans, integration between the overall designer and the developers and designers of the individual components. Thus, one pattern to emerge in product development is this differentiation into development through product architecture and development through humans working together in close coordination.

There is another pattern in product development: namely, differentiation into decentralized coordination through the market (open development pattern) and cooperative coordination via organizations or networks (closed development pattern). Product development that uses the intangible artifacts of product architecture strongly complements open-development pattern product development carried out through the market. This is because the role each component will play in the overall product is clarified in advance. On the other hand, the closed development pattern, in which the tacit knowledge shared within organizations or networks can be protected and incorporated without the use of intellectual property rights, is better suited to complementing the close coordination of developers (see table).

Table: Product architecture and development patterns

Product architecture / development patternOpenClosed
Development standard as artifacts
Overall design carried out first
Desktop PCs (IBM-PC)Production of automobiles by approved drawing system
Developers' close coordination
Component design carried out first (concurrently)
Chinese production of automobiles by mix-and-match of componentsSubnotebook PCs

Semiconductor lithography

The effect on patterns of human cooperation

Present-day product system development is thus influenced at a deep level by the rapid increase in sophistication and complexity of artifacts, but the effect of these increases on cooperation among humans is not just seen in product systems. A good example in the recent evolution of semiconductor production systems is reported by Chuma (2006), who shows that the increased complexity of artifacts substantially affects human-artifact interaction and cooperation among humans.

The present-day semiconductor manufacturing system is highly automated; automation, however, certainly does not mean that human participation will no longer be necessary. The performance of a semiconductor production plant will vary greatly in accordance with the cooperation mechanism installed by the human organization, and this is a major factor in differentiating between competitive superiority and inferiority. As already stated, an artifact's capacity for information processing, no matter how refined and sophisticated the artifact, is restricted to mechanical information processing. It is therefore necessary to create patterns of cooperation that complement mechanical information processing by concentrating on the contextual information processing activities of which only humans are capable.

An important point to consider in the creation of relations of a division of labor and cooperation between artifacts and humans is the fact that present-day, computer-based automation generates useful information regarding production activities. Zuboff (1988) maintains that this fact constitutes the essential difference between 19th century automation and the computerized automation of the 20th century. Accordingly, a structural outline has been formed in which humans are devoted to the information processing of which only they are capable (i.e., problem solving), using information generated by artifacts within production systems.

Chuma (2006), for example, cites the case of Hiroshima Elpida Memory, Inc. In this company, detailed information regarding the production system is frequently compiled in the form of a Work in Progress (WIP) Status Report. This is shared between managers, skilled workers, and engineers, and the task of detecting and problem-solving any incidents that occur within the production system is carried out on a daily basis. The WIP Status Report becomes the basis of a "mutual cognitive environment" (Sperber and Wilson, 1986); additionally, it is a valuable motivator as it clarifies the mission of workers.

Present-day semiconductor production is a complex system involving sophisticated devices in different combinations. There is a coordination problem here similar to the one previously referred to in product system development, and it should be noted that this problem is subject to the phenomenon of increasing complexity. It is difficult to solve this coordination problem solely by relying on adjustments between humans. It is therefore necessary for areas of coordination that can be implemented mechanically to be left as much as possible to artifacts, and for humans to carry out the coordination in areas that cannot be resolved in that way. The Manufacturing Execution System (MES), an intangible artifact that generates the WIP Status Report, plays exactly this sort of role. Moreover, it is of particular interest that the MES provides useful information that then needs to be shared in a form that strongly clarifies the current status, in order to allow humans to devote themselves to troubleshooting any incidents that arise.

The MES plays another, very important role in that by changing the way information is shared within the organization it greatly affects people's incentives. This point has received little attention in conventional economic theory. In cooperation patterns in which human coordination accounts for the main role in organizational activity, people may well use information that they alone possess to their own advantage; thus potentially manipulating it or concealing it from others. As the MES is itself an intangible artifact, it can function as a commitment device that will always disclose required information as shared information.

Other implications of the increased complexity of artifacts

As shown above, the perspective of increasing artifact complexity and the division of labor/cooperation between humans and artifacts affords new ways of looking at product architecture and the way humans cooperate within the corporate structure. These are not the only examples in which this perspective is valid. As the future product systems become increasingly complex, it becomes harder for a single company to encompass all the technology embodied in these systems; thus problem-solving by collaboration between multiple companies becomes unavoidable. The increasing complexity of artifacts is also having a tremendous effect on research activities relating to product systems.

New economic phenomena bring specific areas of economic activity into detail, as if viewed under a magnifying glass; this has the effect of making continually existent phenomena that have attracted little attention the subject of new inquiry. This also stimulates the development of economic theory. From both the perspectives of the increasing complexity of artifacts and that of division of labor and cooperation between humans and artifacts, it is very much hoped that new economic theories will be brought forth.

>> Original text in Japanese

Reference(s)

Chuma, H., 2004. "Increasing Complexity and Limits of Organization Facing Japanese Science-based Industry: from examples in the semiconductor lithography industry," Hitotsubashi Business Review Vol. 52 No. 3, pp. 64-85.

Chuma, H., 2006. "Exploring Factors Behind the Weakening Competitiveness of Japan's Semiconductor Production System: from the viewpoint of meta-level integration capability," RIETI Discussion Paper Series 06-J-043, Research Institute of Economy, Trade and Industry (in Japanese).

Fujimoto, T., 2005. "A Note on Comparative Advantage of Architectures," RIETI Discussion Paper Series 05-J-013, Research Institute of Economy, Trade and Industry (in Japanese).

Nobeoka K., M. Ito, and H. Morita, 2006. "Failure to Capture Value Because of Commoditization - the Case of Digital Home Electronics," RIETI Discussion Paper Series 06-J-017, Research Institute of Economy, Trade and Industry (in Japanese).

Okuno M., H. Takizawa, and Y. Watanabe, 2006. "Increasing Complexity of Artifacts and the Role of Product Architecture," RIETI Discussion Paper Series 06-J-038, Research Institute of Economy, Trade and Industry (in Japanese).

Oshika T. and T. Fujimoto, 2006. "Empirical Analyses of Product Architecture Theory and International Trade Theory," RIETI Discussion Paper Series 06-J-015, Research Institute of Economy, Trade and Industry (in Japanese).

Simon, H., 1996. The Sciences of the Artificial, Karl Taylor Compton Lectures, 3rd ed. MIT Press (Japanese translation: Herbert Simon, 1999. The Sciences of the Artificial, translated by Inaba & Yoshihara, Personal Media).

Sperber, D. and D. Wilson, 1986. Relevance: Communication and Cognition, 2nd ed., Oxford, Blackwell (Japanese translation: Sperber and Wilson, 1999. Relevance: Communication and Cognition, translated by Uchida et al., Kenkyusha).

Suchman, L., 1987. Plans and Situated Actions: the problem of human-machine communication, Cambridge, U.K., Cambridge University Press (Japanese translation: Lucy Suchman, 1999. Plans and Situated Actions: the problem of human-machine communication, translated by Saeki et al., Sangyo-Tosho).

Zuboff, S., 1988. In the Age of the Smart Machine: The Future of Work and Power, N.Y., Basic Books.

February 15, 2006

Article(s) by this author