THE INNOVATION-CO-EVOLUTION-COMPLEXITY PERSPECTIVE
The Internet is a multi-purpose infrastructure for innumerable and highly diverse applications. This considerably limits the prospects of deriving generalized economic consequences.
The identification of and focus on one distinct set of Internet-based innovations - algorithmic selection - is an effort to take a more differentiated look at its economic and social implications. Other analytical challenges are the great significance of technological change and its interplay with economic, political and social transformations.This chapter adopts an integrated innovation-co-evolution-complexity perspective (Latzer, 2013a), an evolutionary economics of innovation (Frenken, 2006), which conceives media change as an innovation-driven, co-evolutionary process in a complex environment, marked by adaptive, non-linear system behavior (Schultze and Whitt, Chapter 3 this volume). Algorithmic selection by search engines and recommendation systems on the micro level, for example, result in unpredictable, unintended emergent effects on the link structure of the WWW at the macro level. Accordingly, the Internet is understood as an open adaptive system, an ‘innovation machine’ because of its specific (end-to-end) architectural design (Whitt and Schultze, 2009; van Schewick, 2010). Co-evolution - sometimes addressed as co-construction or confluence (Benkler, 2006) - is a durable relation between agents that influence each other’s evolutionary paths. Hence, according to a complexity economics perspective (Beinhocker, 2006), processes in economics, politics, technology and society are driven by mutually selective pressure or adaption. This explains the reciprocal interplay - more precisely the pressure and adaptive behavior of technology, organizations and business models that nurture each other. The advantages of such a co-evolutionary perspective include its contribution to better understanding and integrating evolutionary technological change (Ziman, 2000) - where technology is not only output but also input into the economy; to overcome the antagonism of technological and social determinism (Rip, 2007); and to direct the focus from static assessments to dynamic approaches.
Finally, such a co-evolutionary perspective results in other (adaptive) strategies for media management and governance than traditional approaches alone, due to an acknowledgment of the limited predictability and steerability of dynamic co-evolutionary developments. Strategies seek less to dictate developments, avoiding attempts to pick winners from technological alternatives and different business models, and are more oriented to enabling and fostering co-evolutionary processes by creating favorable frameworks, for example by strengthening adaptive policies and feedback mechanisms (Latzer, 2013b, 2014).Selecting and relevance-assigning algorithms on the Internet can be understood, with reference to Bresnahan (2010), as micro general purpose technologies, as widely used clusters of (radical) innovations that enable and trigger innovations in many other economic sectors, because they offer not one specific solution but various new opportunities. The co-evolution with political, economic and cultural factors determines what opportunities will ultimately be used and what the consequences will be for socio-economic welfare. Governance activities to minimize risks - discussed below - are closely interlinked with economic factors and also interact with technological characteristics.
Algorithmic selection can lead to creative destruction, and even has the potential to be a disruptive technology (Christensen, 1997), a special form of creative destruction marked by inferior technology and the replacement of incumbents (low-end disruption, e.g., credit scoring, and new market disruptions, e.g., computational advertising). Innovations are co-evolutionary, adaptive processes of renewal, marked by variation, selection and adaptive reactions. Corporations play a crucial role in selection processes of technologies and of appropriate business models. This will be described in the following sections of this chapter, together with other characteristics of markets of services using algorithmic selection and their market phases.
Starting from an innovation-co-evolution-complexity perspective, several other approaches help to better understand algorithmic selection. The power of technology and the ability of algorithms to shape realities and societies have been variously discussed by researchers and journalists who focus, among other things, on the role of algorithms as agents (Machill and Beiler, 2007), institutions (Napoli, 2013), ideologies (Mager, 2012) and gatekeepers (Jurgens et al., 2011; Wallace, 2016). An institutional point of view, for example, highlights the enabling and restricting role of technologies in general and of algorithms in particular.
Further, algorithmic selection can be conceived as a mode of intermediation (Aguila-Obra et al., 2007), which is central, for example, to understanding platforms and multi-sided markets. It connects supply and demand, that is, providers and consumers of products and content. Algorithms are involved in the allocation of resources, and often have the role of market makers in the value creation system - discussed below. Additionally, the intermediation perspective highlights the role as gatekeeper and its effects on the public sphere and public opinion formation as well as its role in the algorithmic construction of realities that differs considerably from realities constructed by traditional mass media (Just and Latzer, 2016).
19.4