The Triumph of Ethernet

Stanford University Press

Contents

Chapter One

It is a commonplace that the high-tech businesses of Silicon Valley succeed or fail in an atmosphere of intense, unremitting cutthroat competition. But there is more involved in the ways that these firms pursue success than simple competitive rivalry. In the early 1980s, for example, Ronald Crane, an innovative engineer at 3Com who helped that company realize Robert Metcalfe's vision of networking desktop PCs, would frequently assist engineers at competing firms to implement the Ethernet standard in their products. He could do so because he, the engineers he assisted, and the firms for whom they all worked were more than just competitors. They also were members of a technological community.

"We [at 3Com] saw our growth linked to the growth of Ethernet as a whole, and anything that made Ethernet look bad would hurt us," Crane explained. "So, it was in our interest that all of our competitors at least got their interfaces right so that the standards worked out. If the rest of their products failed, that was fine, but we did not want their product to fail because Ethernet failed." As a result of such collaboration and information sharing, the Ethernet community was not just a techno-economic phenomenon based on the competing firms' adherence to a technological standard. It was also a social phenomenon because of the dense web of personal and business interactions that grew up around that standard.

A technological community includes all those firms, organizations, and individuals that are directly or indirectly involved in the development, manufacturing, and distribution of a particular technology or standard. It thus is the social matrix within which a technology or standard is embedded, and the community's economic, social, cultural, and political forces shape the technology's evolution. Depending on the specific technology, the participating organizations may vary significantly, and they may include not just manufacturers, suppliers, and resellers, but also standards-setting bodies, universities, and professional organizations.

A technological community is not just an aggregation of firms, organizations, and individuals attracted by self-interest to the same technology or standard, however. As is the case with other communities, from neighborhoods to small towns to large cities, technological communities are cemented by allegiances, strong or weak, to a common ethos or identity, to common standards or values, and to common ends-allegiances held by people who interact and communicate to develop and benefit from symbiotic relationships. A variety of such communities came into being in the LAN industry. Some were not really communities at all, just aggregations of autonomous firms associated only by virtue of temporarily shared interests in a proprietary or sponsored technology. Others were genuine communities but never became strong ones because they remained small, or because their ties remained weak, or both. In many ways, the paradigm of a strong and vibrant technological community was the one that grew up around a common commitment to the Ethernet standard, a commitment engendered by that standard's open status.

The kind of technology community that the Ethernet community exemplifies exhibits the same characteristics as other kinds of communities. They are far from homogeneous. The mere existence of a community does not ensure Edenic harmony, as anyone who has lived in a small town knows. Competition, of course, is one of the principal motives driving activity within a technological community. Its members compete for the same customers, often with similar products. Although competition is one of the principal sources of a technological community's vitality, it is not the only one.

Technological communities also can display a remarkable amount of collaboration because of their members' commitment to a common standard. Firms, for instance, may collaborate directly on the development of a new component, or they may have reseller agreements. And like many communities, technological communities derive much of their vitality from differentiation and specialization, enabling the community to adapt and survive in ways a more homogeneous aggregation cannot, but also requiring collaboration and cooperation so that the different specialized parts supplied by a variety of firms complement each other and can be assembled into a complete network. This can result either from direct negotiation or as a natural consequence of firms' efforts to seek unoccupied, lucrative market niches.

Not all communities are similarly structured, and this is true of technological communities as well. In general, the larger the community becomes, the more differentiated it becomes, in the same way that large cities tend to have more specialty shops than small towns. In addition, the larger and more differentiated the community, the more competition there is within it. The more firms that offer the same component, for example, the more intense the competition becomes and the more radically prices decline in relation to the prices of comparable components of a different technology. And the larger, more differentiated, and more specialized the community becomes, the greater the variety in the products that it offers.

In technological communities, one of the principal determinants of how the community is structured is the kind of technological standard around which the community is constituted. In particular, communities constituted around an open technology tend to have a different structure than those that grow up around more closely held or proprietary technologies. They tend to be decentralized and nonhierarchical, because no single company controls the technology, and thus no company can exclude other firms from manufacturing the technology and from proposing innovations. Communities that arise around more proprietary technologies are more hierarchically structured, because a single company may exert strong or absolute control over whether and in what manner other firms are allowed to employ the technology.

These differences in community structure have a definite effect on the outcomes of battles to become market standards. As will become clear, less hierarchically structured technological communities often have the advantage over more hierarchically structured communities. In hierarchically structured communities, a single entity regulates the way in which new knowledge and innovations are produced. As a result, different and even conflicting strategies can rarely be pursued simultaneously. In nonhierarchically structured communities, such as those that arise around an open technological standard, by contrast, different, specialized firms pursue different paths, and as a result the technology does not easily become locked in to a declining market.

The mere creation of an open standard, however, does not guarantee that a technology will prevail over more closely held technologies and become the market standard. As will be shown in more detail, like the DIX alliance, IBM turned its Token Ring technology into an open IEEE standard, but unlike Ethernet, Token Ring did not attract a large, vital, nonhierarchically structured community. Because IBM sought to be and to remain the principal player in the Token Ring realm, especially in terms of market share, and thus was the principal competitor of other Token Ring suppliers, the resulting Token Ring community had severe hierarchical characteristics, and there were few incentives for other firms to join it. What proves more important than creating a de facto open standard is creating attractive business opportunities for a large number of suppliers. Open technologies quite often fulfill this prerequisite, but as the case of IBM shows, dominant firms may possess sufficient leverage to violate it, to the eventual detriment of their own dominant position.

Standardization is thus indeed a deeply social process, and in the emergence of Ethernet as the market standard in the LAN industry, as we will see, socioeconomic factors stand out as determinants of standardization battles, especially at their outset. The history of the LAN industry suggests that, all other things being equal, the technology with the broadest and most diversified social support has the best chance to prevail.

All other things are never equal, however. Technical factors, in particular, differ; otherwise there would be no basis for competition between technologies. But in the LAN industry technical factors, in and of themselves, were not ultimately decisive in shaping the outcome of the standards battle. Instead, they were quite malleable, set and readjusted by their sponsors in response to market exigencies. What proved more important than any short-term technical advantage was the relative size and diversity of the technological communities involved. It would be erroneous to conclude that the underlying technology played no role, and that social factors were all that mattered. But the technology proved to be the raw material on which the technological communities worked to achieve the outcomes they desired, not the fundamental basis on which one outcome was determined.

Standardization cannot be understood without examining the role of users. There is no standardization if users do not adopt the technology, and users often provide crucial feedback to the producers, as von Hippel has pointed out. Moreover, users are an important group of actors participating in the social processes that shape the technological outcome. Hence, it seemed only logical to include users in the technological community as well.

In the present community concept, however, users are not included. This exclusion not only follows the convention of the literature but also is justified by the goal of explaining the advantages of open technologies that accrue on the supply side. Besides, the exclusion of users results partly from empirical difficulties in finding original material on users' experiences and locating individuals who can be interviewed. This approach does not seem to be a significant drawback in explaining Ethernet's success, however. As will be described, during the very early standardization efforts Ethernet was primarily championed by its inventor, Robert Metcalfe, and the DIX group. Similarly, users played a negligible role in the de jure standardization by the IEEE. The ones casting the votes at the IEEE were the representatives of the firms producing LAN technology, not the users. Once Ethernet had become a de jure IEEE standard, most users simply followed the LAN recommendation of their computer manufacturers, as this created the least amount of compatibility problems. Since Ethernet had gained support from some of the most powerful computer manufacturers at that time, especially DEC and H-P, it had a clear edge in this regard. In other words, the manufacturers clearly took the end users' preferences into consideration and created a standard. But there was little inherent demand for a technology that possessed the particular characteristics of Ethernet. The end users strongly demanded a standard, but a somewhat different technology could also have solved their networking needs. Consequently, the users granted the manufacturers much leeway in shaping the technology and simply adopted their eventual offering, as long as it suited their needs and as long as it was in fact a standard.

Capabilities and Communities

Many alternative approaches to explaining technological evolution have qualified the initial insights of economic standardization theory and gone beyond the assumptions of neoclassical economics on which those insights were based. Today, a broader conception of a technology's social and institutional embeddedness seems better suited to grasp the processes by which technologies become market standards. The social and geographic dimensions of innovation and standardization have received considerable attention over the past thirty years. Innovating firms often draw from knowledge accumulated by users and suppliers, we now recognize. In the approach stressed by adherents to the paradigm of the social construction of technology, the concept of a "network of actors" is employed to explain what shapes technology. Other scholars have contended that firms often incorporate knowledge that is "stored" in a regional district such as Silicon Valley, whereas others have drawn the geographic line of the knowledge repository even more broadly, suggesting the framework of a "national system of innovation."

Certainly, the regional advantages of Silicon Valley and Boston's Route 128 played an important role in the creation of the LAN industry. In fact, it comes as no surprise that most LAN companies were established in these two areas. However, looking at how people cluster around a particular place, rather than around a particular technology, has the drawback of not permitting us to focus on factors that are clearly relevant to that technology as well. The emphasis on geographic space makes it difficult to compare the effect of social and institutional factors on the evolution of those technologies that reside within the same geographic space.

That is why several researchers have proposed the concept of a technological community as the appropriate way to conceptualize the social dimension of technological innovation and standardization. Edward Constant, for example, argues that turbojet technology was in the possession of and advanced within a "community of practitioners" comprising both individual engineers and organizations like firms. Leonard Lynn, Mohan Reddy, and John Aram have developed the concept of an "innovation community" to link the corporate development process of a technology to its institutional environment. Such a community comprises universities, unions, trade associations, and professional societies. Embarking from a research tradition in population ecology, James Wade claims that "each technology can be thought of as being supported by a community of organizations that have a stake in the technology or design." Although these scholars differ in their definition of a technological community, they all draw the institutional line around a particular technology, instead of a region.

Because the evolution of technology in the LAN industry revolved around issues posed by open and proprietary standards, classifying the technological communities to be found there in those terms makes the most sense. One could of course classify communities otherwise, along dimensions such as the type and numbers of members (that is, the involvement of universities, the military complex, or standards-setting bodies, in addition to firms), the nature of functional integration, the forms of interaction, or geographic scope. But classifying the LAN communities in terms of their relative openness emphasizes what is key to the processes by which each was able to compete and by which the proponents of Ethernet eventually triumphed: the different capabilities possessed by its firms and the advantages that accrued to the technology because its firms possessed those capabilities.

CAPABILITIES

Although the concept of capabilities as a way to understand the ways in which businesses grow-or fail to grow-was originally applied to factors at work within individual firms, the concept has evolved so that it is applicable to the successes and failures of technological communities as well. In a seminal work first published in 1959, The Theory of the Growth of the Firm, Edith Penrose approached the firm not as an abstract production function, as it had been treated in neoclassical theory, but as an administrative, planning organization endowed with real people and a collection of productive resources. According to her, the growth of firms mainly depends on the efforts of managers whose job it is to find opportunities for further growth and to remove obstacles. But managerial efforts may also limit a firm's growth. Although a (growing) firm can hire new managers, they need firm-specific experience and must function as a team to be effective. These requirements cannot be met in a market transaction; instead they must be built through a time-consuming learning process within the firm.

Although Penrose did not use the term "capability"-it was coined in this context by G. B. Richardson in 1972-these claims became the cornerstone of the capability approach to explaining how and why economic growth occurs within firms. The capability approach begins with the notion that the sources of competitive advantage must be built within the firm and cannot be acquired through market transactions. In other words, firms, rather than the environments in which they operate (such as the market), are responsible for their own competitive advantages.

Economists long ignored the ideas of Penrose, but in the 1980s and 1990s various scholars discovered, extended, and refined her work. David Teece, Gary Pisano, and Amy Shuen, in particular, have developed a useful, integrative framework that explains how strategic advantage emerges in this paradigm. In their view, a firm's competitive advantage depends on its capabilities, as well as on its endowment with assets such as specialized plant and equipment, reputation, skilled personnel, financial assets and so on, factors that they call the "asset position" of a firm. Capabilities and asset position, in large measure, determine what a firm is capable of in terms of production, knowledge creation, innovation, and market adjustments.

(Continues...)

Excerpted from The Triumph of Ethernetby Urs von Burg Copyright © 2001 by Board of Trustees of the Leland Stanford Junior University. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.