The Technology: Has the Digital Communication Technology Changed the Way Markets Function? Cooperation or Competition?

Abstract

As social beings, humans have traded goods and services for generations. The mechanics of exchange have adapted over time to the changing environment but the underlying incentives remain the same. Differences in preferences and resources are bridged by connecting with individuals in other parts of the economic network using digital communication technology (DCT). DCT has generated new scarcities and new mechanisms for resource allocation, enlarging the scope of exchange along three dimensions. One is the trend toward organizational restructuring (OR) precipitated by granularity and disintermediation. The second trend is the emergence of organizational behemoths (OB), fostered by network effects. And third, competition for scarce resources is channeled into cooperation as individuals adapt to a rapidly changing context for exchange, and this adaptation itself changes the environment.

Appendix – Graph Theory

The Internet means connectivity, so we need a clear definition of connectedness. Graph theory is a branch of mathematics which provides us with a framework for understanding connectedness [14].⁸

A graph is a representation of a network. It consists of nodes which are connected by links or edges. Nodes represent market participants or economic agents and links represent interactions between them. These interactions could be actual transactions or simply trading arrangements, where some form of economic interaction is, or will be, consummated.

These links are of two types. The first is a link characterized by the logical structure of the network, the basic pattern. This pattern is governed by the logic of belonging to a particular group, a market where nodes trade. The second is a link created by the fact that individual outcomes are impacted by the behavior of all other individuals in the network. In this way, links are governed by behavior.

Links can be formed between nodes because they have a mutual friend (Triadic Closure), they share a common interest (Focal Closure), or they influence friends to join an interest (Membership Closure).

There are many patterns of connections between nodes. A path, Fig. 1.1(a), is a sequence of linked nodes. A cycle, Fig. 1.1(b), is a path where the initial and terminal nodes are identical. Figure 1.1(b) is also a directed graph, where every link has directionality and points from some node A to some other node B, and is also a connected graph, since there is a path between every pair of nodes. A graph is strongly connected if there is an unbroken path between every pair of nodes.

The distance between nodes is the shortest number of hops or links between these nodes.

Components are distinct groups of linked nodes and a cluster is a component with densely connected nodes, as in Fig. 1.1(c). A structural hole is the empty space between two components that may be linked via a bridge.

A gatekeeper node, G, in Fig. 1.1(d) has the property that all nodes in one component need to go through this gatekeeper to get to other components. Consequently, gatekeepers have enormous power over components. Copyrighted technology, such as computer software, can have greater economic significance than other copyrighted material such as books and movies. Computer software often controls the gateway or interface between other software and the copyrighted software in question. Microsoft has copyright protection over the interface between the Windows operating system for desktops and operating systems on servers. Some would argue that search engines, such as Google, have similar gatekeeper status granted by consensus across users.

Node G is pivotal since the shortest path between two components goes through this node.

G-A is a bridge since it connects distinct components and is the only path between these two components; a local bridge is the shortest path between two nodes but not the only path. Since the end nodes of a bridge have gatekeeper status, they have no common neighbors.

Some ratios reveal subtle underlying relationships such as the clustering coefficient of a node A (CC_A) and neighborhood overlap of a link AC (NO_AG). Both CC_A and NO_AG are between zero and one.

The clustering coefficient of node A, or CC _A, is the proportion of total neighbors that are linked. If many of your friends also know each other then you have a high clustering coefficient. For example, if you belong to a group, then many of your friends also belong to this group. Figure 1.1(e) gives examples of a zero clustering coefficient as well as a perfect clustering coefficient of one.

The neighborhood overlap of a link AG, or NO_AG, is the proportion of total neighbors of both A and G that are shared by both nodes. If there are many common friends between A and G, then there is high neighborhood overlap. The number of common friends is called the embeddedness of the link.

In Fig. 1.1(e), suppose A represents Apple on the left diagram and none of its developers know each other, then the clustering coefficient of Apple is zero. On the right diagram in Fig. 1.1(e), if the Apple Watch brings developers together, then Apple’s clustering coefficient is one.

As another example, in Fig. 1.1(d), A and G are friends and A works at Amazon and G works at Google, but none of the other workers at either of these institutions know each other. Then the neighborhood overlap of the link AG is zero or NO_AG = 0. However, both A and G have bridging capital and have the opportunity to create potentially useful connections between these two distinct groups. Now, suppose there is a merger between Amazon and Google, and all colleagues of A and G get to know each other (not shown in diagram). When both A and G share all their neighbors, NO_AG = 1, thus generating bonding capital. We can also say that A and G are in a highly embedded network.

While the clustering coefficient is a powerful concept for nodes, neighborhood overlap addresses connectivity at the level of the network. The two ratios are independent of each other, except in one case. Consider a node A and some neighbor G in the left diagram in Fig. 1.1(e). If CC_A = 0 then NO_AG = 0 because if not then there exist shared neighbors, which implies that two of A’s neighbors are linked which contradicts the assumption that CC_A = 0.

No such implications can be drawn when either ratio is 1 or when neighborhood overlap is zero. For example, in Fig. 1.1(d), NO_AG = 0 but CC_A > 0. This is because CC_A > 0 does not preclude NO_AG = 0. All of A’s neighbors can be linked, except G, maintaining the idea that A and G share no neighbors.

Similarly, again in Fig. 1.1(d), CC_B = 1 does not imply that NO_AB = 1. While all of B’s neighbors are linked, A’s neighbor G is not linked to B. Hence, NO_AB < 1.

Finally, NO_AG = 1 does not imply that CC_A = 1, since shared neighbors themselves may not be linked, so the clustering coefficient could be less than 1. In Fig. 1.1(b), if A and C were linked, they share neighbors B and D, which are unconnected.

Fig. 1.1

(a) Path; graph not connected. (b) Cycle; directed graph; connected graph. (c) Two distinct components; each is k connected cluster; structural hole between them. (d) G is gatekeeper; G is pivotal from path J to B; GA is a bridge; NO_AG = 0 (neighborhood overlap = 0). (e) CC_A =0 CC_A =1; NO_AG = 0; NO_AG = 1. (f) CC_A > 0; NO_AG = 0. (g) Buyers; traders; sellers; [B, C]; [A, G]; [J, K]