Can specialized services be justified?
In an earlier post I wrote about the âawkward periodâ of net neutrality regulation in the EU between the new rules taking effect on April 30th and a crucial explanatory guideline by European regulators group BEREC expected later this year.  No-one really can tell to which degree critical cases like zero-rating and specialized services are still permitted or prohibited (see ars or heise (German)).
In a talk on the future of net neutrality in Europe, Barbara van Schewick listed specialized services as acceptable exceptions to the equal treatment of data if âthese services cannot be provided otherwiseâ. I got the chance later on to ask her, what services she had in mind and she mentioned remote surgical procedures. This unexpectedly echoed a statement by EU Commissioner for Digital Economy and Society GĂźnther Oettinger earlier that day on that very stage, where he told journalist Ingrid Scheithauer specialized services would constitute a narrow but essential part of internet service in the future. With their guaranteed quality they would enable self-driving trucks to drive on our Autobahn with 1.10 Metre distance to each other (a somewhat exceptionally precise example).
I always incorporated an exception for specialized services in my (self-proclaimed) âmoderateâ definition of net neutrality, despite being weary about the missing examples of today. Remote surgery? Really? But after the two remarks on Monday I started to change my mind. My argument boils down to three cornerstones: missing functionality, missing provability and missing authority:
1. Missing functionality A point best illustrated by a chart curtesy of ENISA (h/t to Uta Meier-Hahn):
What do the Internet's basic mechanisms guarantee? So be thankful for every packet. Thx for the reminder, @enisa_eu pic.twitter.com/kmHYJLEVp2
â Uta Meier-Hahn (@zielwasser)
30. April 2016
The Internet Protocol is by its very construction unreliable. It is not suitable for keeping up vitally important systems. If a connected car is not able to drive safely as soon as its internet connection fails, it should not be allowed on public streets. No matter how well built the underlying network and how high the priority of its data connection, if enough cars are driving based on the same connection, queuing of the data packages will delay their transmission. Jitter and lag are the consequences and every other internet based application has to deal with them. It is causing enough harm if computer gaming is interrupted, but no engineer will want to take responsibility for fatal car crashes in meatspace. The same is true for remote surgery or remote control of essential industrial processes.
2. Missing provability For less important functions, how do we know they could not be provided using ânormalâ neutral Internet transfer? Thinking back 10 years, no-one would have expected that best-effort would be a good enough system to transmit high definition video chats from one private home to another, but Niklas ZennstrĂśm and Janus Friis proved everyone wrong in 2003 with their VoIP app Skype. They used their experience from working on peer-to-peer Software and created a decentralized P2P-network that circumvented the otherwise meagre reliability between two private internet connections. Todays most used specialized service is multicast of TV content. Do we with absolute necessity need to jeopardize fundamental principles of the internet architecture to create fast lanes for 4K video? Or could new codecs or fttx connections provide the same benefits without these drawbacks? How do we find out if the category of a specific specialized service is no longer needed after five or ten years? Will ISPs tell content providers themselves, that their services are no longer needed? I simply see no way for a sure-fire proof that a service needs that kind of special treatment.
3. Missing authority Last but not least one question comes to mind: Who would be responsible to decide which services offer societal benefits to such an extend that a special treatment is deemed appropriate. In the current regulatory system of the EU, this decision would fall into the hands of national regulatory agencies (NRAs). NRAs are set up quite differently across the EU, but most do incorporate multiple regulated industries (in the german case post, rail, electricity, gas and telecommunications). Their view is highly technical and did not always incorporate broader cultural and societal questions into the decision making process (donât get me started on the usual revolving door of industry and public service employees). They tend to apply a liberal wait-and-see approach, which is in line with the EUs light touch regulatory approach and can be effective in some areas but produces slow results. Slow means bad, when it comes to fast moving markets of cutting edge internet services. For the sake of a digital european single market, BEREC could be another authority, but the need to bring together multiple national stakeholders makes BEREC even slower. And involving national or regional politics is questionable in itself. Why shouldnât we rely on the global internet governance system that provided us with over 20 years of thriving internet expansion? Oh wait, the IETF does not produce binding decisions, only RFCs. ICANN is itself in a process of massive structural changes, the IGF can only work as long as it does not produce any conclusion.
So there we are. Specialized services are a nice idea in theory, but they 1. donât work, 2. are impossible to be proven as essential as advertised and 3. no one could reliably and with clear authority decide which service is and which is not a âspecialâ service. So in conclusion I have to admit: I canât find any way of justifying to write such a mysterious loophole into net neutrality rules anymore. Please tell me if you find reasons why Iâm wrong. Iâm @ixfel. Iâm apparently against specialized services now.
