Organizational Knowledge-Bases

Authors using this broader concept of organizational knowledge have sometimes invoked one of three broad abstractions to represent an organizational knowledge-base. First, they have used the concept of a set (Nelson and Winter, 1982; Kogut and Zander, 1992, 1996; Nonaka and Takeuchi, 1995; Ahuja and Katila, 2001; Nerkar and Roberts, 2004). A set representation assumes that an organization’s knowledge can be represented as a collection of individual, discrete elements or quanta. A set representation enables the quantification of an organization’s knowledge-base, is broad based and generic, and allows basic algebraic operations to be performed such as union and intersection. All these properties can be usefully exploited in organizational knowledge contexts.

For instance, the cardinal number of a set can be used to serve as an indicator of the size of an organization’s knowledge-base, and can thus be used to describe both the absolute size of a knowledge-base as well as its relative size in a comparison between two knowledge-bases (Ahuja and Katila, 2001). The elements can reflect technical knowledge, knowledge about individual routines or procedures, or even organizational beliefs or heuristics (Nelson and Winter, 1982; Kogut and Zander, 1992; Grant, 1996; Nerkar and Roberts, 2004). Operations such as union and intersection can be meaningfully interpreted in the organizational context as the aggregation of knowledge-bases (for instance through mergers, alliances, acquisitions, franchising) and the degree of overlap or relatedness between two knowledge-bases, respectively (Kogut, 1988; Lyles and Salk, 1996; Simonin,1999; Khanna, Gulati, and Nohria, 1998; Inkpen, 2000; Gupta and Govindarajan, 2000; Ahuja and Katila, 2001; Sorenson and Sorensen, 2001). These properties can also be used to model the processes by which innovation occurs, such as recombination, or to explore the implications of knowledge depth and breadth. For instance, an organization’s depth of knowledge can be conceived of as its frequency of reuse of its existing knowledge elements while breadth of knowledge can be conceived in terms of the number of knowledge elements in the knowledge-base (Katila and Ahuja, 2002).

However, representing a knowledge-base as a set also introduces limitations. First, the notion that all knowledge may be represented as discrete quanta or units may not be a useful simplification in specific contexts. Second, very importantly, the set abstraction assumes that the individual elements are all stand-alone pieces of knowledge, without connection to any other knowledge elements inside or outside the knowledge-base. Yet, most knowledge is usually related to and understood in the context of other knowledge. This limitation suggests that an alternate concept of organizational knowledge that allows the possibility of representing the knowledge elements with some recognition of the links between elements of knowledge may be a superior alternative in some cases. This suggests the possibility of at least two distinct representations incorporating this feature—the organizational knowledge-base represented as a matrix (Grant, 1996; Dyer and Singh, 1998; Helfat and Raubitschek, 2000; Eppinger, Sosa, and Rowles, 2004; Grant and Baden-Fuller, 2004; Siggelkow and Rivkin, 2005; Ethiraj, 2007), or as a network (Hansen, 2002; Reagans and McEvily, 2003; Yayavaram and Ahuja, 2008; Galunic and Rodan, 1998; Almeida and Phene, 2004; Owen-Smith and Powell, 2004; Nerkar and Paruchuri, 2005; Inkpen and Tsang, 2005).

The work on complexity theory and Simon’s seminal work on the design of organizations (Simon, 1957; 1962; 1973) provides the precedent for the matrix representation of knowledge. In its simplest form this entails the listing of all knowledge elements as the two dimensions of a matrix with the actual cells of the matrix being denominated as ones or zeros (Steward, 1981; Smith and Eppinger, 1997) to indicate the presence or absence of a linkage between the knowledge elements. The relation specifying the presence of a one or zero can be defined in many different ways to reflect interdependencies of various types. This matrix can then be used to understand non-obvious relationships between knowledge elements, including potentially by using the power of matrix algebra.

An alternate, but similar, approach to addressing the issue of potential relationships between elements of knowledge is to conceive of the organizational knowledge-base as a network with the nodes representing individual bits of knowledge and the ties representing connections between specific pieces of knowledge. So considered, ties could represent many different forms of relationships. For instance Yayavaram and Ahuja (2008) define a tie between two knowledge elements as a coupling, the decision by an inventor or manager to consider two elements of knowledge jointly thus either using them together or not using them at all. According to them,

couplings thus reflect an organization’s revealed beliefs about which elements of knowledge are most likely to work well together and should be combined and, conversely, what kind of elements are unrelated to each other and do not need to be considered jointly. Couplings can vary in their intensity, going from strong (elements X and Y are always considered together) to weak (X and Y are considered together occasionally) to non-existent (X and Y are always considered independently).

(Yayavaram and Ahuja, 2008)

They then demonstrate the use of coupling and knowledge elements to articulate the knowledge network of individual firms. The network representation of a knowledge-base is fairly general in that, as the above illustration shows, ties can be used to represent a wide variety of dependence relationships and can also be permitted to vary in strength (Ghoshal, Korine, and Szulanski, 1994; Hansen, 2002; Schulz, 2003; Levin and Cross, 2003).

To understand the implications of these various conceptualizations of organizational knowledge-bases it would be useful to have in mind the basic models through which we expect knowledge to create innovations. It is to this task that we turn our attention next. A survey of the literature suggests at least three broad (non-exclusive) processes through which innovations are commonly created—recombinant search, cognitive search, and experiential search (March and Simon, 1958; Cyert and March,1963; Nelson and Winter, 1982; Huber, 1991; Kogut and Zander, 1992; Galunic and Rodan, 1998; Fleming, 2001; Gavetti and Levinthal, 2000; Winter, 2000; Fleming and Sorenson, 2004; Gavetti, Levinthal, and Rivkin, 2005).