David Farber is considered to be the “grandfather of the Internet”, and he gives some valuable insights in the following two video clips as to how the Internet actually works as opposed to common misconception.  This interview is in stark contrast to Vint Cerf’s interview with Cecilia Kang where Cerf claimed that wireless network management wasn’t different from wired networks.  While video is a wonderful medium, it isn’t easily searchable or citable or quickly consumable so I’ve attempted to summarize the Interview below each video clip (not exact quotations).

Part 1 – David Farber interview

(Hover over the links for a pop up box of the video segments.  The videos in full are posted after the break.)

0:00 – Targeting only broadband access providers for regulation is oversimplifying things

0:50 – Traffic shaping and prioritization is normal and justified inside the Internet

1:15 – Regulating just the access portion doesn’t make much sense without considering the Internet as a whole

1:30 – Talks about Akamai and Google operating two of the largest bypass networks i.e., private Internets which are examples of “prioritization, shaping, the whole 9 yards”.  Will these be regulated?  We don’t know because no one has considered the complexity of such regulations.

2:20 – Internet Engineering Task Force (IETF) which is an international organization regulates the Internet.  If IETF comes out with something innovative that can help the whole world, do they need FCC clearance as to whether it is reasonable?  FCC has never had this regulatory power over the IETF for good reason, and they’re not prepared to handle the complexity.

3:15 – The wireless network is different.  Spectrum is expensive, and sold for a particular purpose.  Spectrum usage may be more efficient in future, but we don’t know how to do it now.  Wireless networks can’t be required to treat everything the same way.  Spectrum is scarce and easily consumed.  Wireless network engineering is very different from wired networks even though both carry Internet Protocol (IP) traffic.

4:25 – Internet was created so that very different networks could communicate with each other.  Before the Internet, there were separate wireless and wired development networks with very different approaches.  Bits on a wired network are cheap; bits on a wireless network costs a lot of money.

5:10 – Next generation wireless networks e.g., 4G are not a domestic (regulatory) issue.

5:58 – It was an mistake for the FCC NPRM to get into wireless regulation.  Just because wireless networks like wired networks connect to the Internet, it doesn’t mean that the two networks should be regulated the same.  The Internet supports a wide range of technology including two cups and a string, and we certainly wouldn’t regulate that.

6:26 – Internet was developed and evolved over 40 years, and it is based on electrical signals.  But now we’re going into an era of optical computing.  Intel for example is developing silicon photonics.  We may need a whole new set of protocols designed for optical computing because current protocols may not work at the newer speeds.

8:00 – Last thing we need when people are trying to innovate is to require engineers to go to Washington to ask permission because it slows down innovation in a very fundamental way.

8:20 – Can flexibility be built into FCC regulations?  In theory yes, but FCC isn’t set up to do this nor do they have the technical capability to do so.

Part 2 – David Farber interview

0:00 – Link Hoewing asks about the End-to-end principle where innovation is supposed to happen at the edge of the network and not inside the network.  David Farber refutes this notion that innovation should only happen on the edges.

Note: Richard Bennett wrote a great paper which also debunks this notion.  Farber says End-to-end was a rebellion against the Intelligent network e.g., the Bell network.  Wikipedia also has a great history of End-to-end which debunks the myth of the dumb network and cites many founders of the Internet including an original author of End-to-end who pointed out that “from the beginning, the end to end arguments revolved around requirements that could be implemented correctly at the end-points; if implementation inside the network is the only way to accomplish the requirement, then an end to end argument isn’t appropriate in the first place”.

0:45 – Good case for intelligence inside the network.

1:35 – Look at Cisco routers.  They are very elaborate management systems that are considerably more complicated than just passing packets.

2:05 – Peering agreements on the Internet today are not equitable.  Networks have the right to discriminate and these are commercial agreements between networks.  Will the FCC attempt to regulate peering agreements on the Internet?

2:25 – FCC tried to get access to Internet peering agreements when Farber was chief technologist at FCC to evaluate reliability of the Internet, and FCC was told to “go fish” (though those may not have been the exact words).  These are private industrial agreements.

3:00 – Cloud computing is an old idea.  It is not “end-to-end” and we don’t want to squelch it (with regulation).

3:45 – Did anyone predict the amount of video on the Internet?  Farber says no, and we would be richer than Bill Gates if we did.

5:00 – FCC regulation not needed.  Monitoring and oversight may be good.

5:15 – Nothing wrong with having principles.  The FCC could say that if you step past a certain line, we are going to talk to you.  Not going to take you to court, but we’re going to have a long talk with you.  This strategy has been very effective in the past.  Principles should be guidance and not rules because we don’t know if the rules will hinder the development of the Internet or the next generation of the Internet.

6:00 – Farber thinks people can live with the first four principles.  (I would take issue with the real possibility that the NPRM bans existing service plans or at least opens up torrent of lawsuits with lawyers citing the literal interpretation of the NPRM’s first four principles).

6:10 – Network management stuff needs to be talked about a lot more.  (I agree and the big problem is the prohibition of good network management which would protect low bandwidth real-time applications like VoIP or online gaming).