Internet Pioneers Panel Discusses Challenges Facing the Internet
                    by Ronda Hauben
                    rh120@columbia.edu
 
     The place: Sanders Theater, Harvard University, Cambridge, 
MA. The date: Tuesday, August 31, 1999. The time, 17:15. (1)
 
     The Moderator is Bob Metcalfe, the inventor of ethernet. 
Among the panelists are Louis Pouzin and Hubert Zimmerman who 
created the Cyclades packet switching network in France in the 
early 1970s; Larry Roberts and Len Kleinrock pioneers of the 
ARPANET, the earliest packet switching network; Bob Kahn and Vint 
Cerf, also ARPANET pioneers, who went on to design the TCP/IP 
protocol for the internetworking of diverse packet switching 
networks; and Paul Baran, whose research helped to pioneer the 
development of packet switching technology. 
 
     The occasion: the 10th anniversary of the creation of the 
award honoring lifelong contributions to the field of computer 
communications research which has had an important impact on the 
work of others. On the stage is a panel of those who have won the 
award over the past 10 years.
 
     For his first question, Metcalfe asks what the panelists 
think of the new protocol IPv6 which has been created to replace 
IPv4, the current version of the protocol that makes the Internet 
possible. Surprisingly, almost all the panelists say that they 
see IPv6 as problemmatic. Vint Cerf notes that there isn't any 
pressure from users or vendors to make the change. He adds, 
however, that lots of devices are being planned which will need IP 
numbers and thus justify converting to the new protocol.
 
     Len Kleinrock asks why there has not been any attention 
given in the new protocol version to make future changes easier. 
Sandy Fraser, another of the panelists, points to expenditures by 
big business which he believes will lead to a large installed 
base making it expensive to make future changes.
 
     Bob Kahn reminds the other panelists about how the 
earliest version of the TCP/IP protocol only anticipated that 
there would be a few networks as part of the Internet. They 
thought they would never need addresses for more than 1/2 dozen 
or dozen national networks. But soon the maverick invention of 
the ethernet spawned local area networks, changing the landscape.
In hindsight, Kahn notes, what should be the future is clear, but 
when going forward, it is hard to see ahead. "We are going to get 
it wrong again," he warns, if there isn't adequate thought put 
into what will be needed. Instead of going by the principle 
"ready, aim, fire," when tackling such problems, he observes, 
there's the tendency to "ready, fire, aim" or "fire, aim, 
ready." He suggests the need to get ready first and then to take 
aim at a problem, in order to be able to recognize whether or not 
the correct problem has been identified.
 
     Kahn proposes that the problem which led to the creation of 
IPv6 may have been looked at in a way that is wrong or that there 
was a need for a different approach to the problem. He notes that 
this is an example of where the research community hasn't done a 
good job of thinking through the future and what is needed.
 
     The discussion moves on to what the problem is that causes 
long delays for some in accessing the world wide web. The 
panelists consider whether there are technical causes of the 
delays which the appropriate research efforts could identify.
 
      Changing the focus of the discussion, Metcalfe asks Louis 
Pouzin what his view is of the creation of ICANN. He asks Pouzin 
if ICANN will blow the whole thing apart. 
 
     Pouzin responds, that since he lives in France, he isn't 
sure what the issues are in the United States, but that you can't 
give up on the concerns of people around the world. The task of 
assigning unique IP numbers is not a real problem, Pouzin 
explains. But there has been a warp in handling it at the 
international level. Pouzin asks: What is wanted? Is there a 
desire for a situation whereby a few years from now a number of 
countries will be up in arms and decide after all they could just 
as well organize their own Internet? That it may be problemmatic, 
but that they can handle it among themselves.
 
     Pouzin explains that the ITU is in charge of allocating 
virtual international resources of communication. That such 
sensitive issues must be handled by an international committee. 
There is no other way as they have the experience and the 
relationship and the habits of diplomacy.
 
     Vint Cerf comments that he can't believe he is hearing 
Pouzin say that the ITU is better. Cerf disagrees that the ITU 
would be appropriate to solve the problems.
 
     Pouzin admits that they are difficult problems but that this 
is the way to handle such difficult problems because there are 
so many items that are national obligations that in the end, this 
is how it is done.
 
     Cerf again disagrees, noting that ICANN has had a difficult 
birth but believes that ICANN is needed because industry is 
the only game in town with regard to who can oversee the Internet 
names and numbers.
 
      Another of the panelists, David Farber, interjects his view 
that ICANN got stuck in quick sand. It should have gotten 
an interim board to get the actual board, and only then taken on 
the difficult issues, he advises. First set up the infrastructure 
and then take on the functions. Farber disagrees that ITU's 
process would be appropriate for the Internet.
 
     In response to a question from Metcalfe about some of the 
technical obstacles to the further development of the Internet, 
Len Kleinrock points to feature shock, or the problem for the 
user of absorbing new interfaces that require hours and hours of 
new learning. 
 
     Opposing the tendency toward the "dumbing down" of the 
network, Kahn describes how the potential of the Internet will be 
lost if people can't themselves interact with the Net. He 
proposes that there should be ways to help people learn to become 
programmers so that every citizen would be able to get the 
Internet to do what he or she wants. Kahn also suggests that 
speech understanding research could help by making it possible to 
give computers verbal rather than typed instructions.
 
     Further discussion on the desirability of voice activation 
leads Kleinrock to warn against voice activated agents given the 
difficulty of human precision in understanding what the 
computer has understood and then in being able to control the 
computer.
 
     Farber refers to the fear of people that the computing 
environment will expose everything a person does online to the 
observation of government. Others on the panel recognize this as 
a problem but also that there is the problem of corporations 
using the Internet to gather information on people. Another 
problem presented is the need to gather data to diagnose 
networking problems.
 
     Metcalfe asks if the Internet architecture will continue to 
scale making it possible for many more networks and computers and 
people to be connected. Sandy Fraser warns that the vision and 
the architecture that has made such scaling possible is being 
eroded. He wonders where the leadership will come from to 
continue to sustain the catenet concept, the concept of a 
diversity of networks being able to interconnect and communicate. 
Also Fraser urges the need to reestablish the importance of basic 
concepts that are at the foundation of the Internet, such as the 
datagram.(2)
 
     Another panelist, Paul Green, reminds the audience of the 
two cultures concept introduced by the British writer C. P. Snow. 
Technology, Green proposes, can be used for enslavement or 
empowerment, and there is a need for understanding and exchange 
between those involved with the liberal arts and the sciences. He 
points to the one sided portrayal of technology in George 
Orwell's book "1984". Instead of such a frightening scenario, the 
diffusion of communication can bring international peace and help 
solve the problems of the disadvantaged, Green adds.
 
     Kahn questions how one can determine the issue of the 
ability to scale the Internet from current knowledge. He compares 
this to trying to predict the ability of the world economy to 
scale. Both are hard to predict, he contends, because we don't 
know what will be invented.
 
     Metcalfe asks if there are any silver bullets that will 
solve current problems. Kahn replies that it is much easier to 
build something reliable than to debug problems. He explains that 
it is crucial to have records of what happens on the Internet to 
be able to solve the technical problems. And though this may fly 
in the face of concerns with privacy, it is crucial to come to 
grips with this problem. It isn't possible to keep the Internet 
reliable without keeping certain kinds of records. He explains 
that the telephone system had found a way to monitor the workings 
of the system and so has been able to solve this problem.
 
     The big expense that AT&T had incurred in buying a cable 
company is presented by Sandy Fraser as the motivation for the 
company to encourage customers to buy as much video and audio as 
possible to pay for AT&T's investment.
 
     Others like Dave Clark and Bob Kahn raise the importance of 
exploring what users will want to do using the Internet, rather 
than deciding for users what they will do. 
 
     As his final challenge to the panel, Metcalfe asks what would be 
the most interesting questions to pose to a graduate student 
contemplating research in the data communications field.
 
     Kleinrock proposes studying the field of nomadic computing. 
Also he suggests exploring issues such as: If one gives up all 
privacy, how much security could one achieve on the Internet? 
That the relationship between security and privacy require study. 
Kahn proposes exploring the relationship of design theory with 
engineering practice. That there is the need to do good design 
and to have a way to have measurable results to compare with the 
theory. Paul Green urges students not to take a micro problem, 
but to work on something that they would be proud of, and to make 
it count. 
 
     The panel ends after two and 1/2 hours. The acoustics in 
the Harvard building known as Memorial Hall where the panelists had 
been seated were poor, making it often difficult to hear each 
other or for some in the audience to hear.  Also most had been 
through a tiring day before the panel gathered at 17:15.  
 
     Despite the difficulties, however, something had been 
achieved. Some of the panelists challenged the fads in Internet 
research and development, urging that the problems need more 
effort to be understood. Several of the panelists freely 
disagreed with each other, yet often did so without any hostility 
or animosity. This led to a discussion where different views were 
presented so that the issues could be explored in a broader way 
than often happens in technical conversations. Also the issues 
examined were for the most part either social, or the discussion 
of the technical issues included social concerns or 
considerations. This, too, was quite different from the narrow 
technical discussion that is often proposed as the model for 
technical issues. 
 
     The panel discussion helped to present a view of the field 
of data communication that contributed to the foundation of the 
Internet and to its early development. Including discussion of 
social concerns as a part of the discussion of the field, helps 
to establish the fact that the user is part of the data network 
and the needs and interests and concerns of the user are an area 
to be included in the field of research and study. This then 
presents a glimpse into the future when the user and the 
interests of the user interacting with the hardware and software 
are recognized as a vital part of the Internet. Instead of 
viewing the user as customers or as victims of commercial firms 
vying for market share, users will be viewed as citizens of an 
online collaborative and participatory networking society, or more 
simply as Netizens. 
 
     The panel did not, however, grapple with the most important 
issues of the continued development of the Internet. Such issues 
have to do with the way that, at least in the United States, the 
academic and government and other public or educational forms of 
Internet development have been subsumed within a commercial 
sphere, where any broader vision of the user as netizen or of the 
need to connect all users has been ceded to industry who only 
view users as customers. JCR Licklider who promoted much of the 
early vision for the development of computer networking, 
maintained that network access must be seen as a right, rather 
than as a privilege. This view required that all the population have 
the ability to have access to the developing computer network.(3) 
And that the network be interactive, encouraging the users to 
participate online and in developing it into something that would 
meet the needs and desires of its users. This vision had the user 
participating in creating the ever developing vision for the 
future of the Internet. That is the challenge that users need to 
take up, taking the torch from the pioneers and carrying it 
forward.
 
-----------------
 
(1) The event was the opening session of Sigcomm'99, sponsored by 
the Special Interest Group (SIG) on Data Communication of the 
Association of Computer Machinery (ACM). The conference was held 
from Tuesday, August 31, 1999 through Friday, September 3, 1999. 
See http://www.acm.org/sigcomm/sigcomm99
 
(2) The datagram was one of the early conceptual and technical 
advances which made it possible to have an internet. A datagram 
is a packet containing only source and destination information in 
addition to the data being transported. It doesn't contain 
information about the path for reaching the destination.
 
(3) See for example "The Computer as a Communication Device", 
by JCR Licklider and Robert Taylor, "Science and Technology: for 
the Technical Men in Management" 76 (April 1968): 21-31. Reprinted 
in "In Memoriam: J.C.R. Licklider: 1915-1990", 21-41, Palo Alto, 
Calif. Digital Systems Research Center, 1990. See 
http://memex.org/lick.html and http://www.columbia.edu/~hauben/netbook