When we first reported that Comcast was actively disconnecting BitTorrent seeds, we never expected that it would lead to a FCC hearing, but it did. Let’s hope it’s for the better.
The second half of today’s hearing (pdf link) started with a number of network and technological experts telling us about the Internet, its history, and its makeup. Of main contention was the line between acceptable, and unacceptable traffic management.
Wise things were said, and the panelists made some good points about the unfairness of the traffic management tools that Comcast uses. There was emphasis on the TCP reset, which means that a few seconds after you connect to someone in a BitTorrent swarm, a peer reset message (RST flag) is sent by Comcast and the upload immediately stops.
Richard Bennett (co-inventor of the twisted-pair system for ethernet, and its protocol, 1BASE5) targeted those opposed to any sort of traffic management in his opening statement saying, “if we can’t control network management, we’ll have to shut down the internet”. David Clark, of the MIT computer science lab, opened by saying that ISPs can either see enemies, or they can see partners, and suggesting that right now, they see the former. He, like almost all the panelists, called the current usage of Sandvine technology ‘troubling’, and said that the user should pick the Quality of Service (QoS) level, not an ISP.
Daniel Weitzner, Director of the Massachusetts Institute of Technology Decentralized Information Group summed up bad traffic management with: “Maybe it’s a bit like the old adage about pornography ‘I know it when I see it’. In this case I know what Comcast is doing is in the camp of unreasonable. These are techniques that hackers would use to deny service to any application on the web, very similar in that regard. It might be interesting to hold a panel of security experts to talk about those kind of mechanisms, I’m certainly not one. But, forging data on the internet is probably outside of the realm of reasonable, and any standards body would deem it to be.”
However, one of the most succinct criticisms of Comcast’s actions came from Prof. David Reed, of MIT’s Media Lab, who suggested that any ISP that didn’t follow the standard solutions evolved over the last 30 years should not advertise themselves as an Internet provider, but instead as a company “offering selective access to portions of the net only”, a description many of Comcast’s customers will probably agree with.
The FCC questioner continued the panel discussion, and pointed out that one of the problems might be that there is no actual data on how busy the network was, something that, from his point of view, would be helpful in determining whether the TCP resets are a unreasonable form of network management or not.
One of the panelists (sorry, they all sound the same) immediately replied to this by pointing out that congestion was not important. He compared the TCP reset to a conversation between two people where a third party – who pretends to be one of the persons engaged in the conversation – says “Stop, this conversation is over”. He added: “I find it uncomfortable that someone in the middle is creating a message to you that appears to come from me, I have a lot of trouble with that.”
At the beginning of the hearing FCC chairman Kevin Martin said that they were willing to step in if needed. Let’s hope they will. Feel free to file a comment if you want to let the FCC know what you think of Comcast’s haxxor skills. A video of the hearing will be available within two days.