Edited by Max H. Boisot, Ian C. MacMillan, and Kyeong Seok Han
I just reviewed Boisot et al.'s Collisions and Collaboration, and in that review I discuss Boisot's I-Space in detail. This collection further discusses and applies I-Space.
The book's purpose is "to provide some theoretical perspective on the nature of organizationally relevant knowledge and to indicate the kind of research that might generate empirically testable hypotheses and hence to further the development of a knowledge-based theory of the firm" (p.6). The I-Space is central to that theorizing, meant to provide insight into "the nature of information and knowledge flows in any system" (p.7). The authors assume that "the speed and extent to which information diffuses within a population of agents is a function of how far that information has been structures" and that "information only becomes knowledge if it gets internalized and becomes part of the recipient's expectational structure—that is, if it affects the recipient's belief structure taken as disposition to act" (p.7).
For me, the interesting stuff started in Chapter 4. The authors start by claiming that "How far knowledge gets articulated determines how speedily and extensively it can be shared" (p.109). Although economics has taken the default assumption that markets underlie all other forms (p.109), Boisot et al. argue that "the heterogeneity of organizations, not their homogeneity, had to be taken as a default assumption when analyzing organizational strategies" (p.110). Since organizations are characterized by heterogeneity, articulation and sharing are thus impeded, and so "transaction costs and benefits remain heavily stacked against the kind of articulation of knowledge that would be required to make the choice of markets either ubiquitous or even symmetric with that of organizations" (p.110). So, they conclude, "in the beginning there was the organization" (p.110).
Given this assumption, the authors outline E-Space (epistemic space), with the axes of codification and abstraction. At the most uncodified, concrete corner, we get embodied knowledge; at the opposite corner, we get abstract symbolic knowledge; and in the middle, we get a band of narrative knowledge (p.123). The authors then add the axis of diffusion in order to produce the familiar I-Space—and we now see that embodied knowledge tends to be the least diffused, while abstract symbolic knowledge tends to be the most diffused (p.131). That is, the more well-compacted and abstract knowledge is, the more mobile and fluid it tends to be. Yet at the same time, codification and abstraction involve data losses (p.131).
Boisot et al. claim implications for economics. For our purposes, the most interesting one is that, below a certain threshold of codification and abstraction, a shared context between sender and receiver favors hierarchies rather than markets (p.136).
Chapter 5 continues exploring the nature of organizations. Here, Boisot discusses the evolution of bureaucracies from Gemeinschaft (local, personalized) to Gesellschaft (large, ubiquitous, impersonal) (p.147). Boisot pegs this shift to the printing press, which provided information storage and diffusion (p.148). Similarly, Boisot argues, modern ICTs might favor options other than bureaucratic hierarchies and competitive markets to other options, such as clan-like networks (p.150).
To make this argument, Boisot discusses bureaucracies, markets, fiefs, and clans in the same terms that he discussed them in Collisions and Collaboration. He adds that we can think of three types of complexity:
- Codification: Descriptive complexity
- Abstraction: Computational complexity
- Diffusion: Relational complexity (p.155)
- The diffusion effect: "at any given level of codification and abstraction, more information will reach more people per unit along the diffusion scale than hitherto" (p.160)
- The bandwidth effect: "any given proportion of the population located at some point along the diffusion scale can be reached at a lower level of codification and abstraction—that is, at a higher bandwidth—than hitherto" (p.160).