Why Europe Needs Community Level Network Neutrality
Last week, I led a submission to the European Commission on why Network Neutrality rules are needed at the Community level, a common regulatory framework for the European Union. The full submission and listed authors is available here.
The debates surrounding network neutrality and the open Internet begin with debates on network architecture and network management, but their implications are far-reaching. The Internet is not a direct private link between two end-points but a common pool resource created from the interconnection of tens of thousands of autonomous networks. The value of the Internet, to those who build it, use it, and build on top of it, is not created by any one network or operator, but depends on access to all endpoints being available on a neutral basis to all. If not carefully restrained, the traffic management practices of each inpidual network can influence, fragment, or foreclose the opportunities the Internet provides for innovation, democracy, and free expression.
The question at hand is how the Internet, this immensely valuable resource, is affected by the actions of an individual operator of one of these autonomous networks. Their unilateral decision to manage or manipulate or block specific traffic on their own networks can have far-reaching effects on the whole and on societal benefits from this resource. The open, competitive markets created by the Internet, from a small restaurant offering their menu online to a major content provider, depend on the open architectures, standard protocols and neutral treatment of traffic as implemented on the underlying networks. However, network management techniques are now increasingly used to create and control an alleged scarcity of bandwidth, in the name of network business models that may distort the Internet’s uses. Discrimination against traffic subjects applications, services, and content to the changing and unpredictable interests of network operators, and the governments who regulate them; a future that risks subjecting citizens to ever-more invasive control technologies and interferes with the open, innovative and competitive market on the Internet.
Ensuring that Internet service providers (ISPs), fixed and mobile, are not in a position to unilaterally discriminate between types of traffic or to exploit artificially created scarcity is essential to maintaining an open Internet. Good network management may require choices on how to deal with congestion but the end-user should be able to decide inpidually what choices are suited to his or hers particular needs, not the ISP. Transparency and competition of networks, while they both have roles in protecting consumers, are insufficient compared with a clear European standard preserving neutral and open networks.
In its public consultation on the open Internet and network neutrality in Europe (Questionnaire),[1] the European Commission (Commission) seeks comment on current bottlenecks in Internet networks, problems that can arise related to net neutrality, the applicability to wireless and different regularity approaches to the management of data packets in communications networks including the relationship of network competition and transparency. The Commission seeks to understand the implications of network management, considering that one network operator’s “management” can affect the broader Internet.
The Internet can be best described as a “network of many different networks,”[2] connecting perse users, services, applications, and content. In fact, the Internet consists of over 26,000 autonomous networks (privately and publicly owned) across the world, cooperating in over 70,000 peering agreements to transport data in a universal address space to and from computers used by private subscribers, enterprises, institutions and government.[3]
These networks are built upon a layered and neutral, open, architecture. From a technical standpoint, as the Questionnaire notes: “[t]he end-to-end principle is one of the central design principles of the [I]nternet.”[4] Packets of data transmitted between users across and between networks are transmitted equally on a first-in, first-out, best efforts framework. In this model, traffic is not prioritized or differentiated between and content, applications, and services on the Internet are all treated equally. A neutral network neither promotes nor hinders any particular applications or content, users are able to create, share, and access online content of their choice. Through the Internet created over-the-top of these interconnected and neutral networks producers of content, applications and devices for online use have been able to experiment and innovate with new technologies, given the relatively low barriers to entry to a market for their services.
The challenge at hand is balancing potentially scarce resources, such as the current bandwidth capacity of networks, while maintaining the competitive and innovative Internet space.[5] Within this context we recognize that the value of the Internet ecosystem is created not by summing the value of each inpidual network and Internet connection, but through the global interconnection networking to all endpoints that creates the immensely valuable “over-the-top” marketplace of the Internet, value not “owned” by any inpidual part of the infrastructure.
Moreover, the Internet is characterized by — and benefits from — spillover effects.[6] When the providers of infrastructure cannot charge for all of the value users derive from it, the surplus attracts users who build new value on top of it. Like toll-free roads that open a remote community to new business activity, Internet connectivity can fuel innovation at many levels, including user-to-user collaborations, faster intra-enterprise communication, and new forms of cooperative development.[7]
The networks that make up the Internet consist of hierarchically-structured layers. Heuristics to describe these layers include the seven-layer Open Systems Interconnections (OSI) stack and focus on the different components of the network,[8] or a simplified four-layer model consisting of networked elements, networks, a layer of platforms, contents and applications, and a consumption layer.[9] The fundamental relationships of the layers are the same in both models. The bottom layers refer to physical components, such as wires, routers, switches, or radios for wireless communication. The networked equipment communicates in another layer, through protocols like Internet Protocol (IP) and Transmission Control Protocol (TCP). Higher levels still are applications and services, such as Skype, Bit Torrent, the World Wide Web, and websites. These comments will focus on the relationship between the network layer, consisting of networks and protocols, and the Internet layer, the resulting cloud of content, applications, and services built by the interconnections of networks and users.
As a platform, the network architecture of the Internet creates low barriers to entry for new ideas and innovations. Interconnected through the Internet Protocol (IP) standard, the networking layers transmit data without bias, creating an open Internet layer. The Questionnaire notes that the Internet “empowers citizens and brings a better quality of life through, for example, better health care, safer transport and easier access to public goods. It is no longer just a communication tool. It is an engine for creating more growth and jobs. It is a platform for the delivery of public and private services.”[10] The ability recognized by the Commission that “[s]o far, ‘[I]nternet access has been more or less unrestricted with access to everything the [I]nternet has to offer (content, applications — some of them only available for a fee) — as opposed to a selection of content/applications pre-approved by the ISP” is a direct result of neutral networks supporting the Internet as an open market over-the-top of these interconnected autonomous networks. As will be discussed later, discrimination to traffic at networking layers impact the layers higher up the stack, such as which current and future applications can be used on the network.
The Questionnaire describes the Internet as a “two-sided market (or multi-sided market).” Notably, while the “Internet” describes the series of networks and can enable connectivity between two endpoints, such as an end-user and Amazon.com, an Internet Service Provider does not provide a continuous circuit between the two, but access to the Internet cloud of tens of thousands of networks. Packets of data are often transferred across networks for “perhaps ten, twenty, or even thirty hops between points.”[11] Further, packets do not always take the same route, with traffic routed around congestion throughout the network path.
Existing as a layer above the networks, this market could more aptly be described as a billion-sided market.[12] The Open Internet Coalition, a group of technology companies and public interest organizations, suggests that because there are a billion routers and nodes, the Internet is actually a billion-sided market.[13] Further, nearly every business has a website, and is thus an “Internet company.” This market does not represent a cleanly delineated relationship between “users” and “developers” or “online service providers.” Rather than viewing creation and consumption as uni-directional flow, we see a complex network in which the same nodes are sometimes creators, sometimes consumers — where neutral communications are critical to link them. In a world where personal information is valuable commodity, such as personal information and photos offered on Facebook or data collected from using Google services, the actions of inpidual Internet users contributes heavily to the Internet economy. Our focus should thus be on enabling these end-users to create and exchange, more than on helping the service providers to build “platforms” to intermediate their interactions and transactions.
C. Implications of Non-Neutral Networks for Competition on the Internet
With the implementation of the best-efforts Internet, the network and Internet layers currently exist as different markets. Neutral networks preserve the current competitive, downstream market of the Internet wherein Internet-based content and applications compete horizontally. As Riley and Topolski explain, the “performance of an application or service online as compared to its competitors is determined by the design and engineering of the application or service,”[14] not by preferential treatment of the traffic of the application or service. Network providers also compete horizontally in a separate Internet access market, where the primary commodities are bandwidth and throughput, and offering these products at through a variety of pricing models.
When these networks are neutral, competition of applications and services is insulated from market incentives of the network layer.[15] One user’s decision to use an application or service does not inherently impact the availability of the same application to another user — application use is non-rival. While uses require different amounts of bandwidth, such as watching a high-definition movie versus sending a text-based email, the ISP does not have direct influence on which applications a user may use with their purchased Internet access.
Bandwidth, however, is finite based on existing network capacity, and in practice is typically shared among users on a local neighborhood/aggregation node or backhaul link, or in the interconnection between autonomous networks. This link can be shared among 50 or more residential subscribers, and while each subscriber may have a connection with a theoretical “up-to” speed, their actual throughput speed can be much lower. As a result, common Internet use can result in last-mile congestion when the cumulative bandwidth consumption of all users on the access link exceeds the capacity of the shared link. The resulting congestion can degrade the service experience of users on the link. Unfortunately if bandwidth is rationed through the prioritization of specific types of traffic, ISPs begin vertically integrating their network layer with the Internet layer. “Prioritization forwards higher priority packets ahead of other traffic, and lower priority packets are negatively affected until there are no higher priority packets to send,” Riley and Topolski explain,[16] increasing the quality of service for some applications but limiting the functionality of others. By allocating bandwidth disproportionally to one application or service versus another, ISPs are imposing their judgment of value between the Internet layer and the user. Further, Lennett notes that with prioritization “networks operators are creating inefficiencies that fail to maximize the utility of these networks for their users.”[17]
Riley and Topolski note that prioritization does not affect an uncongested network, and “all packets are forwarded as they arrive, without delays or drops.”[18] Allowing prioritization can incentivize ISPs to create congestion, make bandwidth a rival good and manage scarcity of bandwidth to capture more value generated by the whole Internet. No longer selling Internet access but also packaged content, ISPs will be in the position of power to implement vertical foreclosure on Internet based applications and services and disrupting the current competitive Internet landscape. In light of the Commission’s continued emphasis on competition, this will be a step backwards by raising barriers to entry and undermining the horizontal competition at the Internet layer.[19] Ongoing vertical integration of content and access providers (as shown by Virgin Media and Virgin Internet in the UK efforts to bundle a music subscription service for example)[20] brings a new problem in this environment. Non-neutral access provided by Virgin Internet – which is already experimenting with deep packet inspection – will serve to reduce competition still further in the very shaky and exceptionally uncreative online cultural content market.[21] Discriminatory networks can also harm Internet businesses and Commission goals, such as increasing legal music downloads.[22] For example, prioritization of traffic by PlusNet in the UK reduces implicates legal music services and downloads for PlusNet users.[23]
D. Net Neutrality Affects More Than Networks
The debates that begin with Net Neutrality and the Open Internet affect far more than technical details. As Barbara van Schewick explains, an open architecture is modular, its swappable parts allow for distributed improvement and experimentation. By contrast, “[i]n an integrated architecture, it is usually not possible to make changes to a component that do not trigger changes in the rest of the system.”[24] If network operators are permitted to cross the layers, integrating control of network, applications, and content, that integration reduces the flexibility of the Internet marketplace for would-be innovators at all layers.
Finally and perhaps most critically, the openness of the Internet is essential to freedom of expression for its end-users. The Internet supports methods of communication or organizing such as email and social-networks as well as the ability to easily share media and ideas through video or photo sharing websites and blogs. Supported by the end-to-end principle, end-users are able to define what information they want to access or share, rather than be steered towards specific content sites through agreements with content providers and networks, find certain websites or content subjected to lower tiers, or find content outright blocked.[25] The Internet’s ability to support political discourse and cultural activity creates spillover benefits even for those not actively posting.[26]
Full submission and recommendations for Europe: here.
