Networks are appropriate where the benefits of co-specialization sharing of joint infrastructure and standards  and other network externalities outweigh the costs of network governance and maintenance. Where there are high transaction costs involved in purchasing technology, a network approach may be more appropriate than a market model, and where uncertainty exists a network may be superior to full integration or acquisition. Historically, networks have evolved from long standing business relationships. Any firm will  have a group of partners that it does regular business with – universities, suppliers, distributors, customers and competitors. Over time mutual knowledge and social bonds develop through repeated dealings, increasing trust and reducing transaction costs. Therefore a firm is more likely to buy or sell technology from members of its network. Firms may be able to access the resources of a wide range of other organizations through direct and indirect relationships, involving different channels of communication and degrees of formalization. Typically, this begins with stronger relationship between a firm and a small number of primary suppliers, which share knowledge at the concept development stage.

In many cases organizational linkages can be traced to strong personal relationships between key individuals in each organization. These linkages may subsequently evolve into a full network  of secondary and tertiary suppliers,  each contributing to the development of a sub-system or component technology, but links with these organizations are weaker and filtered by the primary suppliers.

This process is path-dependent in the sense that past relationships between actors increase the likelihood of future relationships, which can lead to inertia and constrain innovation. Indeed much of the early research on networks concentrated on the constraints networks impose on members – for example, preventing the introduction of ‘superior’ technologies or products by controlling supply and distribution network.

Organizational networks have two characteristics that affect the innovation process: activity cycles and instability. The existence of activity cycles and transaction chains creates constraints within a network. Different activities are systematically  related to each other and  through  repetition are combined to form transaction  chains. This repetition of transactions is the basis of efficiency, but systemic interdependencies create constraints to change. For example, the Swiss watch industry was based on long established networks of small firms with expertise in precision mechanical movement but as a result was slow to respond to the threat of electronic watches from Japan.

Similarly, Japan has a long tradition of formal business groups. There are two types of keiretsu, although the two overlap. The vertical type organizes suppliers and distribution outlets hierarchically beneath a large, industry specific manufacturer. These manufacturers are in turn members of keiretsu which consist of a large bank, insurance company, trading company and representatives of all major industrial groups. These inter-industry keiretsu provide a significant internal market for intermediate products. In theory, benefits of membership of a keiretsu include access to low-cost, long term capital, and access to the expertise of firms in related industries. This is particularly important for high technology firms. In practice, research suggests that membership of keiretsu is associated with below average profitability and growth and independent firms like Honda and Sony are often cited as being more innovative than established members of keiretsu. However, the keiretsu may not be the most appropriate unit of analysis, as many newer, less formal clusters of companies have merged in modern Japan.

However, as the role of a network is different for all its members, there will always be reasons to change the network and possibilities to do so.  A network can never be optimal in any generic sense, as there is no single reference point, but it is inherently adaptable. This inherent instability and imperfection means that networks can evolve over time.

More recent research has examined the opportunities that networks might provide for innovation, and the potential to explicitly design or selectively participate in networks for the purpose of innovation that is a path creating rather than path-dependent process. An engineered network,  a nodal firm actively recruits other members to form   a network without the rationale of environmental interdependence or similar interests. Different types of network may present different opportunities for learning.

Competitive dynamics in networks industries:

*System attributes: Incompatible technologies: Compatible across vendors and products.

Custom components and interfaces: Standard  components.

*Firms strategies: Control standards by protecting proprietary knowledge. Shape standards by sharing knowledge with rivals and complementary markets.

*Source of advantage: Economies of scale, customer lock-in. Economies of scope multiple segments.

In a closed network, a company seeks to develop proprietary standards through scale economies and other actions, and thereby look at consumers and get other related companies into its network. In such cases established companies are able to reinforce their positional advantage by adopting new technologies which have implications for compatibility whereas new entrants or existing firms at the periphery of the network will find it extremely difficult to gain a positional  advantage through innovation. Therefore innovations in one technical sub-field may influence some relationships within the network, but not the whole network. Therefore innovation by established firms at the periphery of the network or by new entrants is more common.   Examples include telephony and power generation and distribution.