NNSquad - Network Neutrality Squad

NNSquad Home Page

NNSquad Mailing List Information

 


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ NNSquad ] Re: Even more on Stanford NN hearing (David P. Reed)


From: David P. Reed [dpreed@reed.com]
Subject: Re: [IP] Merccurynews report on Stanford hearing

Since Mr. Bennett was a witness at Harvard, as was I, I have to say that
in my service on the FCC Technological Advisory Council (under Chairman
Powell) and in my interactions with the government, I choose my
arguments carefully, based on the idea that reason and facts, not
insults, will carry the day.  Bennett seems to think that insults will
do better.  I disagree.  I commend those of you who want to understand
Mr. Bennett's views to look at his blog - which is self-explanatory.


________________________________________
From: Richard Bennett [richard@bennett.com]
Subject: Re: [IP] George Ford is getting booed and heckled

As predicted:

In neutrality debate, carriers get blamed for Net's weaknessesBy Richard
Bennett
Article Launched: 04/17/2008 01:35:28 AM PD

I'm afraid Bennett displays a problem common to the current generation of politicians (and to a lesser extent business leaders). He thinks that the mere fact that a court or commission would take the time to even _listen_ to the other side, or ask searching questions, means that the court/commission is prejudiced against his side.

I started to write that this was a disease of NeoConservatives, but in
fact I have seen it on all sides of the spectrum -- it was merely more
noticeable in Neocons because they had power for a while and were trying
to use it to change decades of rulings they considered adverse.

What's more, Martin and the broadband critics have failed to acknowledge
an underlying truth about the Internet: It was originally designed for
the polite society of network engineering professors and students, not
our rough-and-tumble world of large-scale copyright theft and video
file-sharing. And it has design defects - bugs - that make it vulnerable
to overload and abuse.

Worse yet, Bennett _does_ have a point here. The Internet was designed around many of the same assumptions/goals as the ARPANet, its predecessor. Those assumptions included that the users share the goals of the designers, and that the most important goal is to get the packets through. That attacks, if any, would come from the outside -- damage to the physical structure of the network or the systems attached to it.

It had a concept of security, but it still assumed that the operating
system on each host could be trusted to correctly label each packet with
its security level (or that the host would connect through a node that
_could_ be so trusted).

Hence, it had multiple redundancy, so knocking out a single node would
not prevent packets from being sent -- they would just go through a
possibly longer/slower route.

That's why our protocols (not just TCP/IP, but SMTP, NNTP, etc.) have
little or no defense against forged addresses, malware, and other
problems that have plagued us since the Internet went public in 1993,
the year of the endless September.

It was never designed to defend itself against:

 . forged IP addresses
 . single-point or distributed Denial of Service attacks
 . spam
 . viruses, trojan horses, spyware and other types of malware
distributed through email, news, websites, etc.
 . and, yes, the use of otherwise benign mechanisms (news, email,
websites, P2P) to distribute "stolen property" (copyright violations).

This is not to say that Comcast's/Bennett's proposed reaction is a good
idea.  On the contrary, the insertion of forged packets just makes
things worse -- it's one more place where you can't trust that a packet
contains what it says it contains.  Bennett, while calling attention to
the bugs in the Internet's design, attempts to justify Comcast using
such a bug for their own purposes.

We _do_ have a problem.  At some point, somebody needs to sit down and
design a network that is robust in the presence of attacks launched from
its own nodes.  That would include improved authentication of both IP
addresses and backpaths (for mail, news, etc.).  The current attempts to
deal with the problems remind me of the book _Your Inner Fish_.  We are
taking a network that is designed for one purpose, using it for a
different purpose, and then twisting everything to somehow get it to
work in the new way.  And it does work, but not very well.