VoIP SIG call notes - 10/26/2006 - David Isenberg

[Candace Holman <>]

VoIP SIG Conference Call, October 26th 2006

*Attendees*

Candace Holman - Harvard
Deke Kassabian and Shuman - UPenn
Tim Bastion - Penn State
Jeremy George - Yale
Rick Bertelson - University of Minnesota
Rick Carleton - Vanderbilt
Paul Dial - NCAR
Barry Keith - UC Riverside
Loki Jorgenson - Apparent Networks
Ben Teitelbaum - I2
Garret Yoshimi - UHawaii
Leo O'Shea - Boston College
Philippe Sultan - INRIA

*Discussion*

Today's call features David Isenberg, a longtime employee of Bell Labs. 
In 1997 he authored a paper titled "The Rise Of The Stupid Network" 
which argued that the fundamental difference between the Internet and 
the Public Switched Telephone Network was the locus of intelligence. He 
referred to the PSTN as the intelligent network and the Internet as the 
"stupid network". An intelligent network is locally switched for 
specific applications and has its decisions made in a central location 
while a stupid network runs any application, with or without a 
gatekeeper, and has its decisions made at the edges. This paper achieved 
a level of notoriety, which led to his eventual departure from Bell Labs.

This paper predicted several of the problems encountered today in terms 
of network neutrality. The "stupid network" is inherently neutral as it 
didn't care what kind of traffic it contained, and the lack of any 
gatekeepers allowed the rise of all the popular applications of the 
Internet, including streaming audio and video, instant messaging, email, 
and so on. This was great for users, but bad for traditional telecom 
companies, who noticed that they were being taken out of the value chain.

David mentioned that his first indicator that there would be a backlash 
from telecom companies came in 2003, when Dave Dorman spoke at Intel 
about there being no value in being the owner of a network. The 
companies are now painfully aware that they are losing landline 
customers and are being forced to upgrade backbones and access plans 
without the profit margins that they believe they are entitled to. This 
has led to increased technological efforts by companies to inspect 
packets and discriminate based on the type of traffic, which David 
believes to be in violation of the traditional "common carrier" 
designation afforded to telecom companies.

Three recent milestones mentioned by David illustrate this change in the 
concept of common carriage. The first was an FCC review in 2003 which 
made it possible for Verizon to make fiber in local plant to be exempt 
from sharing. The second was the Supreme Court's so-called Brand X 
decision of 2005 which determined that cable modems are information 
services and therefore don't have to be open to competition. The third 
milestone came a few weeks later, when the FCC stated that DSL is 
fundamentally a private information service and the providers don't have 
to share facilities with competitors.

As the telecom companies have realized their new status they have driven 
changes in the law, reinforced by technology, which brings us to the 
current debate about network neutrality. Telecoms have been pushing for 
changes to existing laws, and the new telecommunications bill which has 
passed the US House contains no network neutrality provision. This law 
is currently stalled between the Senate Commerce Committee, which threw 
out the bill without a network neutrality provision, and the full Senate 
which needs 60 votes to cut off a filibuster which has been promised by 
Senators Kerry and Wyden. Senator Frist told the head of the commerce 
committee, Senator Stevens, that he needs to show him 60 votes which 
Stevens has been unable to do thanks in large part to the Save The 
Internet coalition which delivered over a million letters to the Senate 
in favor of network neutrality. David says that the coalition has 
lobbied 27 senators into publicly supporting network neutrality. 
Following the Senate recess, if there are 60 votes in favor of the bill 
it will come to the floor. A secret, Republican-only conference 
committee would meet to reconcile the House and Senate versions of the 
telecommunications act and would likely pass the law without debate. 
Either the bill doesn’t come to the floor at all, or it does and passes 
without a network neutrality provision.

There is still a lot of interest in network neutrality, but there is no 
real driver. Telecom companies want to enter the video market, like 
cable providers, but would require enabling legislation to pass and 
allow city-by-city franchising. If the bill goes to committee and 
becomes law, it would grant this franchise exemption and there would be 
no further impetus to push network neutrality next year. If the law does 
not pass this year, or congress becomes more balanced, then there is a 
chance that network neutrality would be reconsidered next year.

Ben Teitelbaum asks about common carrier status. He feels that it was 
originally a compromise by telecom companies in order to defer 
liability. He asks if they are allowed to discriminate based on traffic 
if they would become liable for disruptions and abuses. David hasn't 
read about this issue in the bills, but believes the house bill contains 
language that absolves carriers of abuses. David gives a brief history 
of common carriage, which dates back to the 1100s and the doctrine of 
public calling, covering blacksmiths, innkeepers, and other providers of 
services. The belief was that if you offered yourself as having a public 
calling, you had to be competent and had to provide your service to all 
comers at a just price, even if you were the only provider in town. In 
the early 1900s a case before the Supreme Court dealt with Western Union 
helping Associated Press reporters to the exclusion of other news 
organizations. The Court stated that there is no explicit statue but 
common law was strong enough that Western Union had to serve all comers.

Candace, playing devil's advocate, asks about the necessity for network 
neutrality. If companies are forbidden from setting up discriminating 
measures, what measures can they use to stop abuse or illegal activity? 
David answers that in every proposal he's read, there has been an 
exception allowing for network management to counter "harm to the 
network" or abuse. He believes that this is a good system, but fears 
that it could be used as a loophole for carriers, and referred to the 
old ATT case against Hush-a-phone's plastic adapter which caused "harm 
to their network" at a time when phones could only be provided by the 
carriers. David says that he favors tiered service plans based entirely 
on throughput, which would adequately deal with heavy consumers of 
bandwidth or others who would like to run servers. Carriers don't like 
these plans as they fear consumer backlash, but he mentions that one 
cable carrier has stated that a small percentage of users drive most of 
their capital expenditure due to heavy use. The CTO of this company says 
that on one hand he'd like to get rid of these users, but on the other 
hand they are the best customers and the future of Internet users. David 
mentions that when Michael
Powell was head of the FCC, he proposed four Internet freedoms, the 
fourth of
which was the freedom to know explicitly what is offered under the service
plan, including blocked ports or limits on access. He notes that the 
current FCC has altered the wording of the fourth freedom so that 
consumers are not necessarily entitled to explicit details of their 
service plans.

Ben says that he also supports usage based pricing, and doesn’t 
understand why carriers are opposed to this. Dave replies that based on 
limited market research, carriers have shown that customers don't like 
these plans. Ben counters that most customers simply don't like 
unpredictable pricing. Scott Bertelsman of the University of Minnesota 
mentions that another former Bell Labs employee in Minnesota has studied 
issues of pricing complexity, and says that with things like cell phones 
this has led to things like carrying over minutes from one month the 
next. David feels that telecoms are being too
conservative, and believes that they want to charge based on favored and 
less favored applications to create pricing bundles that lead to higher 
profit margins instead of looking at solutions that preserve their 
neutrality.

Loki Jorgensen of Apparent Networks suggest that there is a trend for 
telecom companies to push their way up the layer model and own layers 
above what they are traditionally responsible for. For companies like 
Cisco, this means looking for application specific characteristics 
rather than simply moving packets. How this alters the behavior or 
performance of the network is a secondary concern, but he believes they 
don't have much choice as network access is becoming a commodity which 
isn't profitable. David says that comparing the annual revenue of 
network providers to those who provide content shows that they differ by 
orders of magnitude, and the telecom companies aren't being realistic if 
they believe that controlling content is going to add to their bottom 
line. There is very little market pull for video, for example, but all 
the telecom companies are rushing to get involved. This has to do with 
owning the market before it becomes dominant, allowing them to control 
it in the future.

Ben asks about competition for the last mile. He asks David's opinion on 
access being a natural monopoly, and how to either increase competition 
or deal with a natural monopoly. David says that Andrew Odlyzko wrote a 
paper titled "Privacy, Economics and the Internet" in which he explains 
the value of violating network neutrality, and why the phone companies 
need to continue price discrimination in order to exist. As for the 
necessity of network neutrality if there was competition for the last 
mile, he's unsure. He offers some contrarian ideas, such as robust 
competition in the cell phone market with several providers, none of 
whom have opened their networks or even their handsets. It's not a 
forgone conclusion that competition leads to openness. He mentions a 
paper ("Towards an Economic Framework for Network Neutrality 
Regulation") by Barbara Van Schewick, a former student of Lawrence 
Lessig, which discusses that even when competition exists the costs of 
switching service providers is high enough that switching is not worth 
it and competition is effectively absent. Despite this, David would like 
to see more competition, and duopolies aren't enough. Adding more cell 
phone carriers led to price drops and better service plans, as well as 
lower barriers to switching such as number portability. More carriers 
would be nice but does not override the need for network neutrality.

Candace asks what can be said to companies who say that they need to 
discriminate to pay for innovation and infrastructure. David says that 
Internet 2 External Affairs Vice President Gary Bachula testified to 
congress that I2's research results proved that in virtually every case 
adding bandwidth capacity was cheaper and more efficient than adding 
discrimination and QOS mechanisms. David says that if providers claim to 
not be able to afford to add capacity, then they certainly can't afford 
to add discriminatory technology and there must be another reason that 
they want to discriminate based on content.

A final thought from David is prompted from a participant who states 
that the existing telecom companies need to fade away and be replaced by 
new ones who aren't afraid to innovate. David says that the existing 
providers have a culture of optimizing their network, which is how they 
traditionally made money. They are simply not used to the new world of 
abundance and how it affects their bottom line.