Header Ads Widget

Old Rules in New Regulations – Why “Sender Pays” Is a Direct Threat to the Internet

This month, we published a new Internet Impact Brief on “South Korea’s Interconnection Rules”. Our technical analysis concluded that existing rules and newly proposed legislation will slow the country’s digital development and cause a direct threat to the Internet. The analysis focused on mandatory rules around a very particular interconnection regime of business arrangements between telecom operators and Internet service providers referred to as “Sender Pays”. A similar interconnection regime is being re-discussed in Europe, due to the insistence of telecom operators, despite it being analyzed and rejected a few years ago. 

The basic idea of the Internet is a network of independent networks that interconnect to form a shared system of connectivity across all participants. This model has proven its value time and time again over the last decades, and most recently during the COVID-19 pandemic. The voluntary inter-networking arrangements allows network operators to optimize their connectivity with others to meet their customers’ needs. The result has been an efficient and resilient network that is able to evolve to host new applications (like voice calls or gaming), and to deploy innovative services at a global scale. And this happens without prior contracting with everyone in the system. 

It’s quite literally the result of the Internet “way of networking”. 

The interconnection rules in South Korea, and similar proposals that are now re-entering the European debate, are in direct conflict with this Internet model of networking. They amount to an idea of a regulated settlement model where communicating parties are charged for the traffic they exchange. It is proposed by the telecommunication’s operators, because in fact it’s the settlement model they traditionally used for almost a century for their voice traffic business. That model was useful on a completely different era and technologies that pre-date the Internet. 

In the telephone system, callers were charged for occupying a resource (i.e. a “circuit” or “line”) on the network during a certain period of time. If a user called someone on another network the telecom operator billed for the time they occupied on the circuit, and the cost charged to them by the second network for occupying their circuit (a.k.a. “termination fee”). It is also why long-distance calls could get very expensive since the user was billed for occupying circuits on all networks along the call path. The total cost to the user was a function of the regulated termination fees negotiated between the user’s provider and all other phone networks in the system. 

It was a payment model that made sense in the telephone system’s “way of networking”.

However, when applied to the Internet model of networking, this approach makes no sense. First note that the proponents of the sender-pays model flip the logic to no longer being about charging the entity initiating the communication, the user,  but the entity responding, the content provider. This doesn’t even make sense in the old circuit based charging model. More importantly, the Internet is not a circuit-based network, it is packet-based. Its economic model is fundamentally different. 

Today interconnection in the Internet is voluntarily agreed through one of the following forms of business arrangements: 

A network pays another network to carry traffic to all parts of the Internet (transit);

A network pays another network to exchange traffic between their customers (paid peering);

A network establishes a payment-free relationship with another network to exchange traffic between their customers (often called “settlement-free” peering). 

The settlement-free peering model is the most preferred model. Previous studies estimate that more than 99% of Internet peering agreements are informal, and usually settlement-free.

A “sender pays” interconnection framework that effectively mandates paid peering between networks, for example, between telecom networks and content providers, is in direct conflict with the autonomy of networks in the Internet model. By prescribing a set of requirements for participating parties, it constrains the flexibility of networks to negotiate how they interconnect. This interferes with the Internet’s voluntary nature by which independent networks are free to manage their connectivity arrangements according to local needs. This results in inefficient traffic flows, higher costs of data transmission, a more hierarchical and less resilient network topology, and lower quality of services for users in the country.  That is not to say that an access or content provider cannot negotiate a traffic volume based agreement with its content providing network neighbours. It is just to say that those negotiations should not be regulated.

Furthermore, the Internet “way of networking” establishes that once a network has connected to the Internet, it is reachable by any user in the world because it is part of the global Internet. To do this, it only needs to find a network already connected to the Internet, and negotiate how to interconnect with it. This simple model has allowed the Internet to flourish and continually expand by the many kinds of organizations that choose to connect to it. This is how the Internet reached global scale.  

The proponents of reinserting those old rules into new regulations want content providers to negotiate different arrangements under regulated conditions with many other networks far away around the globe, before they can exchange traffic with them. This raises a strong barrier to be effectively connected to the Internet. The consequence is a form of Internet fragmentation where end-users can only access online services that have contracted with their ISP or telecom provider. And at the quality and conditions stipulated by these arrangements. In addition, and depending on implementation, these proposals are close to charging ‘valuable’ services more than others. The expectation that all packets are the same and therefore treated neutrally, is then broken. 

If we stop treating the Internet as a technology-neutral, general purpose network, we’ll lose it. 

The European Commission and the Council are listening to telecommunication operators in these requests and have chosen to introduce this debate as part of the Declaration on Digital Rights and Principles. Reviving an old debate might be productive, but the arguments have to be backed up by solid technical arguments. We encourage policy makers, especially the signatories to the recent Declaration on the Future of the Internet, to carefully consider the consequences of these type of “sender pays” rules, and to conduct Internet impact assessments with an eye to the implications for the Internet’s networking model. 


Post a Comment