Paul Penfield, Jr., The Internet, in "Plenary Panel: Multi-Network Networks -- Opportunities and Challenges," Executive ComForum: Strategic Directions to the Information Future, International Engineering Consortium; Miami, FL; November 13 - 15, 1997.

The Internet

Paul Penfield, Jr.



... used by the speaker during the talk.

The Internet

Paul Penfield, Jr.
Professor of Electrical Engineering
Head, Department of Electrical Engineering and Computer Science
Massachusetts Institute of Technology
Cambridge, MA 02139-4307
(617) 253-4601

Title Slide

Thank you for the opportunity to participate in this panel discussion.

What is the Internet?

A network of networks, of different types
IP (Internet Protocol)
Can run on many types of communications hardware
   TV Cable (coax)
Can even run on circuit switched networks
Packet Switching
The distinction from circuit switching less important than it once was

Slide 1

The Internet is for data, not voice.

The use of circuit switching is inefficient because of reserved resources that might otherwise be statistically shared.

Philosophy of the Internet

Keep it simple
Keep it general
Delivery guarantees
Best effort
End-point hardware
Specializes the net to particular tasks
Handles errors
Everybody welcome
Multiple types of traffic coexist
Cyberspace is still a frontier, not yet civilized
Business models
Hard to understand how anyone makes money
Fee structure is efficient -- fees aligned with costs

Slide 2

Simple and general mean CHEAP.

Fee structure is efficient in part because it is free to users.

Nothing is free, of course; the Internet has zero variable cost but relatively high fixed cost.

Applications of the Internet

Original idea, from ARPA
Remote login
File Transfer
Cooperative computing
(Robustness was used as argument to justify funding)
Users had other ideas (killer apps)
   Freedom of the Press is no longer only for those who own one
Software distribution
Future technologies and applications
Point-to-point real time (telephony, video)
Specialized data types

Slide 3

Another future technology I forgot to include, FAX


Resource Allocations
IANA (IP addresses; now 32 bits, going to 128 bits)
Network Solutions, Inc. (domain names)
Relaxed Attitude
Based on cooperation, respect, and trust
Aversion to regulation
International aspects
National boundaries permeable
Are traditional political units still relevant?

Slide 4

The standards bodies really do work. The general high level of trust and cooperation, and general lack of commercial competition have induced participation by very highly qualified people.

IETF handles the Internet; the Web Consortium W3C handles the Web. There is participation in both bodies by all the major players.

The governance so far is informal -- no regulation to speak of. This has been possible in part because the Internet has been small potatoes. It remains to be seen how long this lack of regulation will persist.

Serious Issues

Intellectual Property
Domain Names
Navigation, Resource discovery
Rational business model
Importance to society
This week Judge Zobel was let down

Slide 5

Tell them about the Judge's problem, caused by a power failure. It was someone else's industry, not ours.

War is too important to be left to the Generals

The Internet is too important to be left to the engineers.
But then, who?
Telephone companies never saw it coming
   They spend most of their income billing the customers
   They can't even provide U.S. national wireless service
Cable companies have the biggest pipe into the home
   Not robust (my cable modem locks up every couple of days)
Politicians don't have a clue
Entertainment industry thinks only of one-to-many
Equipment suppliers know only what hardware is needed
FCC afraid of the industry (e.g. they allowed HDTV interlace)
Nations increasingly irrelevant
Businessmen don't understand importance of standards
Research establishment has best vision
   They started it, and still know it the best
   But they couldn't manage their way out of a paper bag
Fortunately, no management seems to be necessary!

Slide 6

I've got a lot of cheap shots here -- an insult for everybody in this room.

The issue is whom to trust to provide leadership in getting the Internet widely deployed in homes. You have to bring IP, probably at a rate of several hundred kilobits/second, preferably 1.5 Mb/s, into the home. That is the goal. Anything less will not serve the public well.

About the phone companies, I seriously think copper loops to residential subscribers couldn't run at the needed rate without being too close to the Shannon limit. There is not enough slack.

My personal preference is the cable companies. They have the needed bandwidth. They are having startup problems, and there is a question whether they have the resolve to make the investment. They do not have a lot of capital behind them.

When all you have is a hammer, everything looks like a nail. Just as the telephone companies think of a switched analog service and try to map the Internet onto that, so the entertainment industry sees only passive movies. Their idea of an infrastructure is Cinema 1-2-3-4-5-6-7-8-9-10-infinity.

We have seen examples where commercial interests have undercut standards in order to achieve a proprietary differentiation. Just recall what IBM tried to do to ASCII. Today, IP is accepted as THE digital transmission protocol standard. Any plans to deploy competing standards fail to serve the public interest.

In this slide I should have included the computer industry. Their idea of reliability is not having to reboot more often than once a day.

Now that I have insulted everyone in the room, I will insult my own academic, research community.

The only thing that points to success is the observation that the Internet has gotten along very well, thank you, without regulation, commercial direction, or central planning. The creators of the Web did not have to ask anybody's permission to use the Internet, or to develop a new set of protocols -- they just went ahead a did it.

So my vision for a winning infrastructure, from an engineering point of view: The standard is IP, supported by all computer manufacturers already. Use fiber to get to the neighborhood; coax to a point inside the house; broadband wireless, perhaps carrier current over the power lines, inside the house. At the head end, avoid going through any switch. The customer's computer is always connected. This infrastructure will let a thousand flowers bloom.

Thank you.

URL of this page:
Copyright (c) 1997 Massachusetts Institute of Technology.
Created: Nov 12, 1997  |  Modified: Dec 31, 1998
Related page: Penfield publication list
Site map  |  To Paul Penfield's home page  |  Your comments are welcome.