Those who want to transform the internet into an online service where content providers pay to play are sacrificing the principle that we owe to the likes of Wikipedia, social networks, and the free communication space of the internet as a whole. Seven Theses on net neutrality and diversity by NRW Media Secretary Marc Jan Eumann and Konrad Lischka.
(Published under CC BY-ND 3.0 DE)
A server containing the almost complete Netflix video library looks rather unspectacular: encased in grey metal housing, weighing 45 kilos, 58 centimetres wide, and equipped with 72 hard drives of 3 terabytes each. US-based online streaming video service Netflix provides these systems free of charge to companies that give Netflix lots of traffic. The Netflix boxes symbolize the way digital videos have changed the network infrastructure: content providers are trying to place their servers with cached content as close as possible to their end customers. Netflix and Google do it on their own, smaller media providers (such as German TV networks ARD and ZDF) outsource this to companies like Akamai. Akamai has 150,000 servers housed in 1,200 provider networks in 92 countries around the world.
Why? The longer the route a data packet has to take over the internet and the more intermediate stations it passes, the more error-prone and slower the transmission will be. This is not well received by users, especially by those who are trying to stream video only to be faced with jerky images and artifacts disrupting the picture.
While companies around the world are paying billions for faster, more reliable delivery of their services online, the Council of the European Union is negotiating over net neutrality. Is that a contradiction? Is net neutrality as a principle at risk in the face of these changes? It’s time to clarify the arguments in the neutrality debate. Seven theses on this matter:
1. The termination monopoly among end users remains, which is why open networks are important
From a media policy perspective, delivery services like Akamai have not one key fact: content providers can deliver their content via many networks. But on the last section of the route to the end user, packages from online libraries, wikis and on-demand movie portals need to pass through the network of the user’s ISP. It is this last stretch where there is no competition from alternative routes. Normal internet users cannot select provider X for their media libraries and provider Y for their email. Households typically have only one internet connection and can therefore make no direct comparison of the delivery provided by other ISPs. As a result, it is unlikely that someone will change ISPs just because YouTube videos are jerky.
The goal of net neutrality is an open network. It needs to be open for the widest possible range of media and services for users to select from and not for ISPs to select based on its key position.
2. The total infrastructure of the free internet is a reserve for diversity, not specific content
Media diversity online is inextricably linked to a particular feature of the original network architecture: the basic protocols (TCP/IP) follow the so-called end-to-end principle. This original internet architecture is neutral. Transmitters of data packets cannot exclude individual applications at the end points of the network. Each vendor can reach each recipient over the internet. In addition, each receiver can also be a provider. There is no central authority that controls innovation and use.
This innovation-promoting architecture is what has led to the diversity of content we see on the web. Centrally managed online services such as Minitel, Compuserve and BTX no longer exist. The free web has, however, spawned whole new forms of media where there is no strict separation of roles between senders and receivers, some even without commercial intent, such as Wikipedia, countless urban and special wikis, attention distribution channels such as the Usenet, Slashdot, Reddit or Twitter. And makers of web videos and bloggers who are always bringing up new issues not covered in traditional mass media.
When it comes to media policy, the free communication space that is the internet is particularly relevant for shaping opinion, because diversity is secured by the neutral, end-to-end principle of the web. Ensuring diversity online has to start with an open infrastructure as a whole, not by making final selections of predefined services.
3. Open networks do not contradict differentiated offerings
An open network on the path to the end customer does not prevent ISPs from differentiating the access they offer. If customer have free choice, they must also be able to choose a service from their ISP. For the open Internet, for example, higher bandwidth, unlimited data volume, fastpath. Such additional offers from an ISP do not restrict diversity, if they apply to all content. Growing numbers of VDSL and cable broadband customers suggest that people are paying more money for a better connection to use the free Internet. The innovation architecture of the open network increases the attractiveness of faster access.
In an open network, there is freedom of choice, not egalitarianism. It is not the operators of the infrastructure, but customers and content providers who determine the transmission quality and pay for it. A volume tariff is not problematic in and of itself. But if the infrastructure provider determines that individual websites are excluded from this volume, this restricts the openness and diversity of the internet.
4. The EU Parliament’s proposal is a good basis for ISPs and media
Lawyer Tim Wu developed the concept of net neutrality in the early years of the new century. His principle was that retail ISPs should generally have no influence on the content transmitted through their networks. Intervention would be permitted only in clearly predefined exceptions. Even before the boom in online videos, Wu recognized that content such as videos might represent a special case. In 2005, Wu wrote that services in which packet losses or delays lead to drastic losses in quality may be at a disadvantage in a content-neutral architecture over other services.
Wu suggested that further exceptions to the principle of net neutrality should be permitted if this does not affect the quality of open internet access. To avoid such an impact, exceptions would only be permitted in private networks. Wu presented IP television services (IPTV) as a concrete example. Such services already exist. For example, the Deutsche Telekom’s T-Entertain fits Wu’s definition of an IPTV.
The text adopted by the European Parliament in April 2014 concerning net neutrality takes Wu’s definition further. It stipulates a clear ban on discrimination in the provision of internet access, with some narrowly defined exceptions. The text describes net neutrality as end-user rights from ISPs. The Parliament follows Wu’s distinction between the provision of internet access and so-called special services.
The approach is right: an open internet and closed special services can co-exist in a way that benefits the common good. This depends on the special services not come at the expense of the free internet. Access to the internet is “free” if people can view videos in media libraries or documentation on Vimeo in the necessary quality. Anyone with network access of 10 Mbps or higher should be able to watch a Vimeo video if no other services are being accessed at the same time. The end user must be able to decide whether a certain bandwidth should be reserved for IPTV.
The quality of free access to the internet should continue to grow and access to the internet must not be marginalized to push people to book special services in certain quality classes.
5. New business models are important for ISPs and critical for bilateral markets for content.
The definitions of internet access provision and special services in the text adopted by the European Parliament provides enough room for a balance of interests. It protects the internet’s role as an open space for communication and provides space for the ISPs’ new services.
Many of the ideas mentioned by ISPs are consistent with the principles of an open network:
IPTV services such as T-Entertain meet the definition of a special service: the virtualized network is segregated from the internet, the circle of users is restricted, the service does not replicate existing services on the open internet, but instead offers something new. If a successful special service like does not breach the requirements formulated by the European Parliament, this speaks for the commercial prospects of the ISP.
Some ISPs have suggested that such special services might include applications in the fields of e-health, IP telephony, and video conferencing. In the discussion of the regulation, such concrete examples should be included.
More problematic would be any attempt to create bilateral markets. An end customer with termination monopoly could require content providers to make payments for a specific transmission quality. The end customers would pay for the internet access while the content providers would also pay for the transmission of the content to their end customers. In a worst-case scenario, such a market would take away a consumer’s right to choose the quality of their service, instead placing the decision in the hands of ISPs and content providers.
Online content providers already pay for the delivery of their data in a guaranteed quality. But there is a difference between payments to delivery services such as Akamai and any payments to end-customer ISPs: there are options when it comes to choosing a delivery service. But on the last leg of the journey to the end customer, the data must pass through the network of the customer’s ISP. Due to the termination monopoly, agreements between content providers and ISPs need to be evaluated differently in terms of their impact on media diversity.
6. IP interconnection and peering are relevant topics for media policy.
The boom in online videos has made the debate over open networks more complicated. A kind of data transmission to date almost never discussed in public is becoming increasingly important: how does the data actually get to consumers’ ISPs? Content providers and ISPs in the United States are currently haggling over the details of this transfer.
For example, in February 2014, Netflix signed an agreement with Comcast, a leading ISP in the US. Netflix pays Comcast for better access to the latter’s customers. Before the contract was signed, the Netflix transfer rates had fallen dramatically on Comcast, they have since improved enormously, the problem being the data transfer point. But the companies continue to argue over who was to blame.
For media users and content providers, the quality of data transfer is as important as transmission quality within an ISP’s network. Because the effect is the same if contents are difficult to reach due to lack of capacity or due to traffic management in the ISP’s network.
The retail ISP’s termination monopoly has its full effect in IP interconnection. Peering and the placement of servers with cached content in provider networks are the technically superior method for transmitting video data. And with these interconnections, content providers have no alternatives to retail ISPs. A customer is always only accessible via a network. Therefore, IP interconnections like net neutrality are issues of media diversity. William B. Norton, author of the Internet Peering Playbook, the standard reference work in the industry, says: “When you deliver video, there is in practical terms no alternative but to buy paid peering from retail ISPs.”
The market for IP interconnection is not transparent. When what is paid, why some transfers stutter, where the problems between Netflix and Comcast come from, these can’t be readily understood by outsiders. These outsiders also include end customers. If the video is jerky, customers today can’t understand why.
We need more transparency in IP interconnections as the basis for our future discussions. We can then argue whether and when it is right for ISPs to demand payments for peering. And how is that to be evaluated by CDNs, content delivery networks that deliver data to ISPs on behalf of content providers for a fee? End users might find helpful a user-friendly monitoring tool that shows the exact location of transmission problems in the network between them and the content providers.
7. Separate the expansion debate and diversity issues.
One argument that’s heard again and again in discussions on net neutrality: more and more data is being transferred, which benefits content providers, so they should have to share in the costs of expanding the available bandwidth.
This argument is not entirely wrong, but not detailed enough to address. One has to differentiate: the expansion of which part of the path are content providers supposed to co-finance? Content providers already pay to transfer data across the internet until it reaches the portal of the retail ISPs and that’s a good thing. It’s not clear if higher fees are necessary here. There hasn’t been any talk of severe shortages or lack of investment on these parts of the internet.
The debate on network expansion in certain regions concerns ISP infrastructure. Should content providers have to pay for this? Should they pay for a transmission quality that the customers also have to buy from ISP? Would it increase the common good if, despite the termination monopoly, there would be a move away from the internet’s end-to-end principle? Whether the network access is open, whether a central authority controls innovation, these are primarily questions concerning ISPs’ business model.
It should be clear what kind of network one wants to expand before discussing how to go about it. The open infrastructure of the free internet is now part and parcel of human communication. It will not be possible to sacrifice this open infrastructure in order to expand it in poorly served areas. To put it bluntly: those who want to transform the internet into an online service where content providers pay to play are sacrificing the principle that we owe to the likes of Wikipedia, social networks, and the free communication space of the internet as a whole.