The global policy movement to improve connectivity and close the digital divide has spawned reports on international markets for data and their associated policies. The papers present sometimes confusing terminology: usage, transit, peering, and interconnection. Each term has a specific meaning and practice, which could vary in countries and context. Policymakers could benefit with a summary of the policies and proposed instruments—in addition to a thorough review of their own country’s networks and practices. Here are some key findings from the reports.
The Rise of a Parallel, Proprietary, and Unregulated Internet by Platforms
Germany’s Federal Network Agency commissioned a study on competition in transit and peering markets (141 pages), noting that the issue hasn’t been examined by European regulators for at least 5 years. The report observes that internet traffic in Europe is growing 25 percent year over year, that 80 percent of this traffic video, social media, and games, and that just 5-6 players (e.g. the platforms Netflix, Amazon Prime, YouTube etc.) account for more than half of all traffic. These players have more international backbone capacity than the world’s broadband providers and have jettisoned third party transit in lieu of building their own backbones, undersea cables, and data centers- The transit business has declined as a result. The platforms largely avoid internet exchanges where prices are transparent, instead building bespoke networks for their proprietary content and maximizing the efficiency and profitability of their services.
The massive development and expansion of backbone and delivery infrastructures by these players has permanently changed the overall global architecture of the internet, the structure of interconnection, and the relationship between platforms and broadband providers, creating competitive disadvantages for operators. The sustained growth of internet traffic continues to shape the dynamics of the internet’s architecture, with the continued disproportionate growth of video streaming and cloud services having the greatest impact. Private network provision has many advantages, though conflicts can emerge when parties exchange data, particularly with the market power disparities between the entities. While the internet architecture has changed drastically in the last decade or so, the legal and regulatory framework for traffic flows has changed little, and the largest platforms are essentially unregulated in these international data markets. The exception is South Korea with its unique approach to broadband policy and recognized global leadership in broadband.
Network Usage vs. Termination
South Korea has had a framework for compensation of network usage for almost a decade. The policy ethos reflects a recognition of shared responsibility between broadband providers and content/application providers to ensure the quality of data delivery and user experience. In practice, policy ensures cost recovery of the installation and maintenance of fiber from the content provider to the broadband provider’s core router. This provides dedicated bandwidth for the given content and protects against the degradation of the network experience for users which don’t access that particular content.
Importantly, this practice has nothing to do with termination of the traffic to end users. It appears that Analysys Mason, Internet Society, and others confuse network usage (which describes the relationship between a broadband providers and content/application providers) with the termination regime of “sending party network pays “ (SPNP).
SPNP only applies to Tier 1 broadband providers in South Korea (Korea Telcom, SK Broadband, and LG Uplus). If the traffic exchange ratio between the two operators is 1:1.8, they will pay the same amount of interconnection fee. If the ratio between the two operators is 1:2, then broadband provider A pays the amount of traffic to broadban provider B areceives amount of traff 1.2 (1+ 1- 0.8) from broadband provider B. The traffic fee corresponding to 0.8 exceeding 1 is exempted.
SPNP does not apply between content/application providers and broadband providers.
While cost recovery is encouraged in South Korea, it is not mandatory, and large US players have gamed the system. For example, Netflix rejected claims for cost recovery and brought a broadband provider to court, saying it had no obligation to pay for the broadband network upgrades required to manage Netflix content which increased 26-fold near overnight. Netflix lost, and the case is on appeal.
Similarly Facebook demanded South Korean broadband providers install Facebook servers inside their networks for free. The broadband providers balked; after all, the servers have cost and cannot be repurposed for other content, and hence are inefficient and redundant if housed for free. To force the issue, Facebook shut off some of these servers and re-routed traffic to other countries and operators. This degraded the end user experience, and Korea’s telecom regulator fined Facebook for what it considered as intentional harm. Facebook brought the issue to court and won, but the abuse caught the attention of the Korean Assembly.
The Facebook incident served as an opportunity to introduce the Korean Service Stabilization Act. The act imposes an obligation on the very largest content providers to consult in advance with the broadband provider to prevent user damage if the very large content provider changes the traffic transmission path and suddenly puts traffic burden on the specific broadband provider’s network.
Going forward, the Assembly considers updating the Telecommunications Business Act to deter these problems of refusal to supply from US tech companies, to stipulate that companies engage in a good faith negotiation, and to provide information for traffic data and pricing transparency. The bill has no fee mandates.
Datasets needed for verification
Policymakers have little data about international data markets. While helpful information on international data traffic is available at the aggregate, global level from Cisco and Sandvine, it tells us little about the behavior of the actors within a traffic exchange and the microeconomics of individual networks.
Preliminary efforts are underway to provide more data, notably from Strand Consult which collects data on streaming video data on rural broadband networks and documents the pros and cons of different methodological approaches. Importantly Congress has considered addressing this through the Funding Affordable Internet with Reliable Contributions Act or the FAIR Contributions Act which would empower the FCC to conduct the needed study.
While no system is perfect, there is no systematic data that demonstrates harm from South Korea’s traffic data policy. On the contrary, the country is celebrated for the highest rates of penetration for fiber to the home (86 percent) and 5G (47 percent adoption). It is a first mover in network innovation and a global force in content development for local consumption and export. Moreover Google and Netflix enjoyed a year of record profits in the country. It seems that broadband fair cost recovery goes together with a flourishing ecosystem.