Entrevista con Vint Cerf

abr 18, 12 Entrevista con Vint Cerf

Hace algunos días, tuvimos la suerte y gusto de compartir unos correos con Vinton Cerf, uno de losllamados padres de la Internet quien, durante los años setenta y junto con Bob Kahn, diseñaron lo que hoy conocemos como TCP/IP, la base arquitectónica de la Internet.

Durante noviembre de 2011, tuvimos la oportunidad de conocer a Cerf en su visita a Santiago de Chile, compartimos algunas palabras respecto al desarrollo tecnológico en la región y, luego, cada quien siguió su camino.

En el marco de la elaboración de un paper sobre Neutralidad de la Red, contactamos a Cerf para conocer su postura respecto a algunos temas contingentes, y tenemos el honor de poder hacer pública esta entrevista para ustedes.

A continuación, les dejamos la cadena de correos para futuras referencias.

 

  • De: José Huerta
  • Para: Vint Cerf

Hello Mr. Cerf,

My name is Pepe Huerta, we met briefly in your visit to Santiago, Chile in 2011, during your meeting with the President of the Senate (Senator Girardi).

I dont expect you to remember me, but we are part of the team that made the chilean Net Neutrality bill possible during 2007 to 2010. At the moment we are part of NGO META, focused in civil rights and public policy in telecommunications.

We’re currently ‘trying’ to get a paper published and distributed among general public, our representatives and government authorities regarding the enforcement of our net neutrality law, so I’m also trying to steal some of your time to ask you a few questions regarding TCP/IP history and, if you allow me, quote you in this paper which I gladly will submit for your revision and accuracy check.

I’ve read professor Tim Wu’s “Master Switch: Rise and fall of information empires” (2010), who quotes you (page 209) about the concept of encapsulation and the “envelope” as a design principle for TCP/IP.

My two humble questions regarding this subject are:

1. Creating TCP/IP as a “common language” for public networks, kind of a “wrapping” for privately generated content (as described by professor Wu), does that makes information independent from the infrastructure that carries it?

2. In your opinion, does this independence from infrastructure (if you agree on professor Wu’s vision) makes all privately generated data ‘untouchable’ by the infrastructure holders (Backbones, ISPs, etc.)?

The context for this questions is our approach to explain why QoS or general public network management are, by principle, against network neutrality, even when our own law does allow ISPs to manage the network without interfering with free competition rules.

I appreciate your time and thank you very much for your kind words during your visit to our country.

Kind Regards,


  • De: Vint Cerf
  • Para: José Huerta

On Thu, Apr 12, 2012 at 12:12 AM, Pepe Huerta Estrada wrote:

Hello Mr. Cerf,

My name is Pepe Huerta, we met briefly in your visit to Santiago, Chile in 2011, during your meeting with the President of the Senate (Senator Girardi).

I dont expect you to remember me, but we are part of the team that made the chilean Net Neutrality bill possible during 2007 to 2010. At the moment we are part of NGO META, focused in civil rights and public policy in telecommunications.

We’re currently ‘trying’ to get a paper published and distributed among general public, our representatives and government authorities regarding the enforcement of our net neutrality law, so I’m also trying to steal some of your time to ask you a few questions regarding TCP/IP history and, if you allow me, quote you in this paper which I gladly will submit for your revision and accuracy check.

I’ve read professor Tim Wu’s “Master Switch: Rise and fall of information empires” (2010), who quotes you (page 209) about the concept of encapsulation and the “envelope” as a design principle for TCP/IP.

My two humble questions regarding this subject are:

1. Creating TCP/IP as a “common language” for public networks, kind of a “wrapping” for privately generated content (as described by professor Wu), does that makes information independent from the infrastructure that carries it?

yes, your perceptive observation is the key to the generality of the Internet design. It is NOT a purpose-built network (unlike the telephone system that was built primarily to switch voice/sound). The consequence is that anything you can digitize, you can send over TCP/IP.

2. In your opinion, does this independence from infrastructure (if you agree on professor Wu’s vision) makes all privately generated data ‘untouchable’ by the infrastructure holders (Backbones, ISPs, etc.)?

I think it should be and you can enforce this by end-to-end encryption.

The context for this questions is our approach to explain why QoS or general public network management are, by principle, against network neutrality, even when our own law does allow ISPs to manage the network without interfering with free competition rules.

the QOS argument has been debated for years. Personally, I think increasing network capacity solves most QOS problems but one could allow the user to select a quality of service without revealing the nature of content. In fact, the IP packets have markers for “speed”, “reliability” and “low delay” although, to be honest, these bits are not used much because their interpretation is not uniform across all the networks of the Internet.

I appreciate your time and thank you very much for your kind words during your visit to our country.

Kind Regards,


  • De: José Huerta
  • Para: Vint Cerf

El jueves 12 de abril de 2012 a las 8:15, Vint Cerf escribió:
2. In your opinion, does this independence from infrastructure (if you agree on professor Wu’s vision) makes all privately generated data ‘untouchable’ by the infrastructure holders (Backbones, ISPs, etc.)?

I think it should be and you can enforce this by end-to-end encryption.

But, as a general principle, infrastructure holders can’t manage IP packets as an ordinary person can’t make use (or manage) of private property even if that property isn’t secured or protected?

I personally see this as an ownership problem: who owns each individual part of the communication that’s made possible by the Internet?. The content of each package (in the abstract) is owned by its creator (The Cerfs, the Huertas, the Googles or the Facebooks), land lines, satellite links, routers, servers… are owned by backbones or ISPs, but the actual “envelope”, TCP itself isn’t owned by any particular person, which makes this management something like a force driven activity more than legitimate owner activity.

So, even when end-to-end encryption will solve most of this problems (not sure if this encryption can bypass Network Management or QoS practices), should Governments enforce this natural neutrality of TCP at this level, entering the “force driven” business of trying to keep ISPs from messing with each package as they don’t have any property or ownership rights over them?

Thank you very much Mr. Cerf! I really appreciate this.


  • De: Vint Cerf
  • Para: José Huerta

ISPs and other infrastructure providers have a right to manage traffic to protect resources; they should not be able to use their facilities for anti-competitive purposes. Even with encrypted traffic, the infrastructure provider has to be able to deal with routing and flow or congestion control. What they do not have is a “right” to the ocntent unless it is produced by them or under their auspices. The terms and conditions of production would have to give them rights, before an infrastructure provider could lay claim to content.

vint


  • De: José Huerta
  • Para: Vint Cerf

But in this case, where ISPs have the right to deal with congestion (which you said could be solved by increasing network capacity, subject that opens a whole new discussion about cost efficiency), they tend to create anti-competitive scenarios mostly against their own competition, but also and even more common, between content providers. For example, http direct file download services versus p2p file download services, being the second injured by ISPs anti-congestion measures and unable to compete with the first since their traffic is being shaped at will by the ISP.

Recently we had a publicly criticized case where Telefónica (spanish ISP and backbone that operates in Chile) published their network management policy, according to mandate on the net neutrality law. They openly stated that Megaupload’s traffic is being capped during peak hours. They also pointed megaupload as “a protocol” (not a service), meaning that all http direct download services are being managed for anti-congestion, wich again opens a theorical discussion about the actual capability of an ISP to control traffic for the whole universe of a particular type of content providers.

There are two injured parties in this scheme: Users which can’t access or have a lower quality access to content; and also content providers who can’t operate because ISPs are applying measures to protect their business model.

So, protection against anti-competitive purposes is a wanted good, but it can’t be restricted only to ISPs business, it must protect also content providers against artificial anti-congestion measures that certainly affects their business.

Do you think this is a valid approach to judge network management policies and Government’s role in protection of network neutrality?


  • De: Vint Cerf
  • Para: José Huerta

i think it is important to assure that network management strategies are not biased for competitive reasons. Comcast was shutting off BitTorrent based on protocol. They have since changed tactics and, in times of congestion, they meter traffic according to the traffic class subscribers are paying for.

v


  • De: José Huerta
  • Para: Vint Cerf

Thank you very much Mr. Cerf. I really enjoyed this e-mail exchange and is really helpful for our work. Do I have permission to quote you in our paper?

Thank you very much again for your time!
Kind Regards


  • De: Vint Cerf
  • Para: José Huerta

yes, feel free to quote.