by Vishal Misra (@vishalmisra)
Much has been said about TRAI’s recent ruling (pdf) on differential pricing. Lawyers who depend on nuance don’t like it, mathematicians/engineers who prefer precision are hailing the simple “no differential pricing except for emergency services” ruling. The former group is calling it a sledgehammer, when they would’ve preferred a nuance laden scalpel. One person who belongs to both groups (in that she is both a lawyer and a computer scientist), Barbara van Schewick, was quoted by TRAI on the harms of unclear regulations and case by case adjudication. So contrary to what Marc Andreessen was trying to say before he unfortunately ventured into anti-colonialism, this regulation is in fact anti License Raj.
In this post I want to focus on the other exception that is mentioned in the ruling, and that is the one for private networks. A lot of people are worried that this is a loophole that can be exploited by telcos as a roundabout way of implementing differential pricing. I wouldn’t be so worried, and that’s because of a carefully crafted (one can even say with surgical precision) definition, 2(c) in the ruling, of private networks. Here is the relevant definition
This definition immediately rules out conventional or carrier grade NATs, where devices share a public IP address behind a NAT router. In that model data is indeed transmitted over the (publicly switched) Internet. Carriers can of course host services entirely on their network and create a special connection (virtual appears to be OK) to customer premises. They would need special VPN services installed on all equipment on their network, and install separate VPN connections on consumer devices. Then there is the question of what services they can host on this private network. Since no data can be transmitted over the Internet, we would need content providers to host their content completely on this closed telco network.
Netflix/Amazon/Snapdeal/Flipkart for example would need to host their entire search catalog inside telco premises. And for e-commerce fulfillment, the transaction has to be completed within this closed network as well.
All messaging services would be between customers of the telco. Even a translation via a gateway to the open Internet would violate definition 2(c). You instantly limit the value of messaging services when you limit who you can talk to.
Tech giants like Google and Facebook generate almost their entire revenue via ads and the content economy runs largely on ads. So complete ad-networks and their infrastructure would need to be hosted inside telco premises. At the very least, a call from customer->telco premise->ad-network->telco premise->customer would need to be done. Even if we assume this translation does not violate 2(c), these companies are spending billions every year trying to shave off the microseconds from their response times. I am sure they will love this architecture.
Then, since data is not allowed to escape the private networks, where would the clicks from the ads go to? Only to content that is hosted on the private network — that would be incredibly limiting for the ad-networks. If the clicks have to go out on the public Internet, then the operating systems on the end devices would need to understand that there are two networks they are multi homed on, and even if you manage this seamlessly, the ads would need to open up on a separate “service”. You will need a careful orchestration on the consumer end-devices to ensure things don’t leak between services.
All this seems a lot of work for people of my tribe, the engineers, to pull off. Maybe the telcos can focus on better service as a customer acquisition tactic, and devote engineering resources there.
13 words in 2(c) define private networks and effectively rule out nuancing.
The Scalpel and the Sledgehammer.
Crossposted with permission from the author. Read the original post here.
Vishal Misra is a faculty in the Computer Science Department of Columbia University, where he has been looking at the issue of Internet Economics and Net Neutrality for a number of years. He is also a serial entrepreneur, being part of the founding team of Cricinfo and is now the founder and chief scientist of the data center storage startup Infinio.