DEVELOPMENT AND UNIQUE ECONOMIC ATTRIBUTES OF THE INTERNET

Important economic characteristics of the Internet developed interdependently with its technological basis, its organization, and the level of adoption (see also Chapter 26 by Garcia, this volume).

2.2.1 The Emergence of the Commercial Internet

The first commercial providers of public Internet access emerged in the late 1980s. In the USA, The World and PSINet started operations in 1989 but were initially not allowed to access many university and government installations. This situation changed when an agreement with the National Science Foundation (NSF) was reached.⁴ The development by Tim Berners-Lee and his collaborators of the components of the World Wide Web (WWW) - including the Hypertext Transfer Protocol (HTTP), the Hypertext Markup Language (HTML) and a first web browser - were additional steps toward moving the Internet beyond a specialized community of researchers. Adoption accelerated with the release of the graphical user interface-based browser Mosaic in 1993 and Netscape Navigator in 1994.

Early commercial online networks offered services to closed user groups and employed proprietary technology. CompuServe (founded in 1969) built its initial customer base around the time-sharing of computing resources.

After a short period during which such access was based on measured hourly prices, America Online and other Internet service providers (ISPs) introduced flat monthly rates that further facilitated adoption. Because the US telephone network was organized as a common carrier and flat local telephone pricing was widely adopted by subscribers, it was relatively easy for ISPs to offer dial-up Internet access as a service configured on top of the existing, ubiquitously available telephone network. At peak, nearly 6000 ISPs offered dial-up service and within a few years Internet access via the phone network spread rapidly (see the excellent discussion in Greenstein, 2015).

2.2.2 Migration Toward Broadband

The migration from narrowband to broadband access started in the 1990s and enabled vast innovations in application services in a virtuous cycle of improvements in the network and in applications. Upgrades from narrowband to access speeds above 256 kbps are often referred to as first-generation broadband. Subsequent generations of technol­ogy that can support access speeds up to the gigabit range are commonly labeled as ultra­broadband or next-generation networks (NGNs). First-generation broadband could be deployed by upgrading existing networks but the more advanced next-generation networks require more significant upgrade investment and the increasing rollout of new fiber-optic networks in combination with advanced wireless technology that supports access speeds up to the gigabit range.

Network and service upgrade patterns vary greatly between countries and regions and show strong forms of path dependence. In countries where investment in the commercial Internet is the outcome of entrepreneurial decisions, multiple alternative upgrading paths are visible, leading to different technology mixes in first- and second-generation broadband. In the USA, cable TV service providers seeking to mitigate the competi­tive pressures of a saturated multi-channel video market introduced faster broadband Internet access during the second half of the 1990s. US telephone companies initially invested only slowly in digital subscriber loop/line (DSL) technology but accelerated their investment in response to the rollout strategy of cable systems and regulatory changes that reclassified DSL and mobile Internet access as information services, creating regulatory parity with cable (Bauer, 2005). In response to concerns about preserving an open Internet this trend toward less regulation was reversed in February 2015 when the Federal Communications Commission (FCC) again reclassified all broadband Internet access, independently of the technical platform, as common carrier services (FCC, 2015). Whereas there are many concerns about this approach, it offers for the first time a sym­metric regulatory model. In countries where cable TV did not have such a strong early market position, such as in many European Union member states, regulators retained unbundling rules for broadband. This often resulted in xDSL (all types of DSL lines) remaining the leading platform for broadband Internet access (Cave and Shortall, 2011). In some Asian countries, such as Japan, South Korea, and Singapore, fiber deployment was prioritized by public policy.

The gradual migration to fixed and wireless broadband enabled increasingly higher access speeds and went hand in hand with declines in the price per unit of data for users.⁵ Together with advances in computing, signal compression, and dramatic declines in the costs of information storage, these developments unleashed a broad range of new applications and services and further altered the economic characteristics of the Internet toward a layered architecture and vertically related markets.

2.2.3 Evolving Network Topology and Convergence

Thus, within a relatively short period of time the Internet evolved from the ARPANET of the 1960s to the narrowband Internet in the 1970s and 1980s, the subsequent migra­tion towards broadband Internet from 1990s onwards, and the ongoing convergence to an all-IP platform, in which multiple technologies are seamlessly integrated by logical protocols, especially TCP/IP (Transmission Control Protocol/Internet Protocol). Despite this transformation from a government-funded to a largely commercially funded network, many stakeholders retained the original vision of the Internet as a commons in which traffic capacity is a non-rival good. However, the migration from narrowband to broadband Internet and the different funding models have resulted in a reassessment of the paradigm of universal connectivity using a best-effort network. Congestion and asymmetric traffic flows as well as increasing demand for traffic quality differentiation resulted in the development of more complex interconnection arrangements between carriers, such as asymmetric and paid peering arrangements in which payments are exchanged for traffic volumes exceeding certain thresholds (Besen et al., 2001; Laffont et al., 2001; Faratin et al., 2008).

In this process the mix of technologies employed in access, middle mile, and back­bone networks changed driven by the relative costs and capabilities of different network platforms. Dial-up and first-generation broadband technology essentially utilized the existing networks. The topology of the telephone, cable, and wireless networks had been optimized based on the relative costs of transmission, switching/routing and storage (e.g., Egan, 1996).

The migration to broadband and ultra-broadband further changed the economics of the Internet. First, incremental network upgrades were increasingly insufficient to keep pace with the growing demand for capacity. Rather, new network capacity had to be deployed in backbone and access networks at much higher costs than the migration from dial-up to first-generation DSL or the rollout of first-generation cable modem service. Creating proper incentives for network infrastructure investment became a major chal­lenge for national regulatory agencies, who had designed unbundling and open access rules for environments in which networks had already been installed (Cambini and Jiang, 2009; Gruber and Koutroumpis, 2013; Briglauer et al., 2013). Second, services and appli­cations configured over-the-top (OTT) of the network, such as Skype or Netflix, and increasingly heterogeneous user demand required a broader range of QoS support.

These changes accelerated the decade-long process of convergence between tele­communications, computing, and media content. Digital multi-purpose networks with broadband and ultra-broadband capacity allowed the integration of formerly separated networks and services on an all-IP network so that network operators can use fixed and wireless technologies in complementary ways to grant users seamless access independ­ently of device and location. While all-IP networks are not necessarily the most efficient technical solution, changing user demand and habits, such as the expectation to get access to content anytime and anywhere, added additional momentum to these develop- ments.⁶ These technical and economic developments contributed to a new wave of inno­vation in the core network infrastructure in addition to the vibrant innovation activity on the edges of the network. Furthermore, the changing economic structure contributed to strong pressures toward industry consolidation and vertical integration.

2.2.4 Modularity, Layering and the End-to-end Principle

The Internet is built around a particular set of design principles that influence its eco­nomic attributes and organization. Of particular importance are modularity, layering, and the end-to-end principle. Modular design of technical systems seeks to reduce the interdependence of components so that they are only loosely coupled (Saltzer et al., 1984; Yoo, 2013). It allows individual components to be designed, produced, and possibly used independently (van Schewick, 2010, p. 39). Modularity facilitates certain types of innova­tion processes, as individual modules may be changed without having to alter the entire system. This is very visible in the app economy, where innovative software can be con­figured to reside on the edges of the network, thus avoiding time-consuming and poten­tially costly changes in the core of the network. All that needs to be known to an app developer is the application programming interface (API) to the operating system and the network. Interoperability and standardization are therefore critical for modular systems to work well. Modular innovation processes will often flourish in a competitive market environment (Baldwin and Clark, 2000; Langlois, 2002; Bourreau et al., 2007). However, not all socially beneficial innovations are modular. For example, the first smartphones required considerable efforts to coordinate the value network that could not be delivered by decentralized markets (Ehrlich et al., 2010). Coordination across modules may be dif­ficult in systems that are governed in a decentralized fashion so that more systemic types of innovation may be complicated to achieve. Coupled innovations, such as smartphones or smart grids, in which applications and service innovations are contingent on network innovations (and vice versa), may require other forms of coordination such as exclusive contracts or forms of joint ventures and alliances.

One widely recognized attribute of the Internet related to the modular nature of infor­mation technology is its layered architecture. Layering is a hierarchical technological design feature. The specific setup can be largely explained by its historical evolution and the technical choices made by the engineers involved in its early design but it also turned out to be a robust and scalable solution. Authors have delineated these layers differently. Moreover, the dynamic evolution of the Internet may change them (claffy and Clark, 2014). The most basic distinction is between the physical network layer (link layer), logical layers allowing traffic to seamlessly flow over these heterogeneous networks (IP layer, and transport layer) and application and services layers configured on the former two (Lemley and Lessig, 2001, Yoo, 2013, pp. 1742). Others have devised more finely differentiated layer models (e.g., Whitt, 2007; Fransman, 2010). In this vertical division of labor the Internet layer serves as an integrative layer that homogenizes differences at the physical link layer, where fiber, copper, coaxial cable, wireless, and satellite networks are used to transport signals. Engineering conventions in the Internet allow for higher layers to make use of lower layers down to the Internet layer (but not below that layer). Lower layers are not allowed to make use of higher layers. Thus, the Internet layer func­tions as a portability layer within a framework of relaxed layering (van Schewick, 2010, p. 88). The end-to-end design is related to these features. Expressed in various ways, it is a convention that places commonly used functionality in the logical core of the Internet whereas other functions, applications, and services are configured at the logical edges and in higher layers of the network (see van Schewick, Chapter 14 in this volume for a detailed discussion).

2.2.5 Multi-sided Platforms and Plasticity

Two related bodies of research on technological and economic platforms focus on these pervasive interdependencies. The management literature looks at platforms as ‘technological foundations upon which other products, services, and systems are built’ (Gawer and Cusumano, 2002). This technological perspective is related to but not identi­cal to definitions that emphasize the economic features of platforms as linking different sides of a market (Rysman, 2009). A technology-centric view starts from the realization that important high-tech industries, including computing and telecommunications, have adopted modular technological architectures since the 1960s to facilitate coordination between increasing numbers of components. Platforms enable the assembly of comple­mentary modules into the systems needed to create value. In a complex socio-technical system such as the Internet, multiple platforms co-exist, often arranged in a nested way. Semiconductors enable operating systems, which in turn function as platforms for appli­cations and services. Physical communication networks serve as platforms for informa­tion transportation services, which in turn enable applications and services. Likewise, mobile devices can be seen as platforms that enable mobile data communications and the multitude of applications and services built on them. Some platforms are sufficiently flex­ible to support a wide range of complementary modular technologies and services. Hence they are examples of general purpose technologies (GPTs) (Bresnahan and Trajtenberg, 1995; Helpman, 1998) with a high generative potential (Zittrain, 2008).

Economic models of platform markets recognize these characteristic technological features but broaden the perspective. Hagiu and Wright (2011, p. 1) provide a particularly compelling definition of multi-sided platforms (MSPs), and view them as ‘an organization that creates value primarily by enabling direct interactions between two (or more) distinct types of affiliated customers’. Although there is some variation in what are considered their key economic attributes, all contributors conceptualize platform markets (often also referred to as two- or multi-sided markets) as special type of intermediation. Early contributions emphasized the presence of direct and indirect network effects between the market sides (Rochet and Tirole, 2006; Armstrong, 2006; Evans and Schmalensee, 2007) but more recent papers point out that such network effects, while often present, are not a necessary condition for an intermediary to be a platform (Hagiu and Wright, 2011). One important economic function of platforms is to reduce transaction costs between partici­pants in different market sides. Where externalities are present, platforms can facilitate internalizing them. In this sense, platforms are institutional arrangements to help over­come forms of market failure and obstacles to market transactions.

Taken together, modularity and flexible layering have greatly increased the plasticity and generativity of digital production technology. Plasticity allows the production of digital services and applications with multiple factor combinations at often radically dif­ferent costs (e.g., online video via best-effort Internet connections, peer-to-peer commu­nications [P2P], and CDNs). Within limits, most products and services can be produced with alternative technologies but in a digital environment in which many products and services are software-based the flexibility abounds. The resulting greater plasticity allows new, non-traditional players to enter the market swiftly and hence may significantly alter patterns of competition in digital and traditional markets. A recent example is the rapid intrusion of WhatsApp (owned by Facebook) in the market for voice and messaging services. Generativity refers to the greatly expanded space of innovation opportunities opened up by digital technology and the accelerated pace at which it is being explored. It also has direct consequences for the intensity and dynamics of competition.

2.3