This evening I "attended" the 2005 A.M. Turing Lecture which was broadcast live over the internet from the University of Pennsylvania.  The UIUC Computer Science department was kind enough to put it on a big screen in an auditorium.  But they were also unkind enough not to advertise it at all to the broader university community.  Thanks for the heads up Jenny!

Assessing the Internet: Lessons Learned, Strategies for
Evolution, and Future Possibilities

Vinton Cerf and Robert Kahn, winners of the 2004 A. M. Turing Award "For pioneering work on internetworking, including the design and
implementation of the Internet’s basic communications protocols,
TCP/IP, and for inspired leadership in networking" "are two key pioneers of
"internet"-working (dating back to 1973), and ACM is particularly
pleased to invite "all users of the Internet" to attend this year’s ACM
Turing Lecture."  Moderated by Lyman Chapin.

"A Protocol for Packet Network Interconnection," the title of the 1973 paper that started it all; distributed at a special meeting of the INWG at Sussex University in September, 1973, and then finalized and published in the IEEE Transactions of Communications Technology, in May, 1974.
20 Years—One Standard: The Story of TCP/IP
Vinton Cerf [some Real Audio interviews]
Robert Kahn

I will post what notes I was able to take and hope that they’ll make some sense to anyone interested.

1973 Internetworked ARPANet, packet radio and packet satellite
1974 "A Protocol for Packet Networking Interconnection" published
1978 The protocol was split into 2 parts: TCP and IP

Reminded that the internet = the logical framework.  Often today, this is forgotten and the internet is equated with the hardware that implements it.  But that implementation is not the key, as it can be "multiply realized" (my comment).  The logical framework is the key. [Kahn]

Layers, "gateways" linking the nets (now called routers), self-configured (routing protocols) [Cerf]

Layering led to effective implementation strategies [Kahn]

Assumptions were built-in; allows for evolution of technical and social structures
Social structures are a key element.  Approx. 1983 the community began taking some repsonsibility (IAB, IETF, and others). [Kahn]

Security was thought of from the beginning.  But the work was classified so couldn’t be discussed with the rapidly widening community. [Cerf]

"The edge of the net" – for Cerf this is where the IP layer stops and host computers start.  But then the next question is which edge?
"The edge" has spawned enormous creativity. [Cerf]

What has been some of features of the internet that has helped generate so much innovation?

A1 – The layers are stable, thus can have massive innovation in implementing each layer.
A2 – As long as "off the edge" protocols agree most anything can be done across the net; e.g., P2P [Cerf]

The internet protocol was designed to run on any transmission/switching protocol; thus, Cerf’s famous slogan (and t-shirt) "IP on everything." [Cerf]

One could say that the internet is "infinitely incrementally evolvable." [Chapin]

"Yes," but:
  It is hard to add mobility (at many levels)
  Multi-homing
  ID management/authentication [Kahn]

Because, assumptions of architecture may have inherent limits [Cerf]

One initial assumption – Computers are usually connected to the net; rarely do they disconnect.  Not even close to the case today.

Persistent identifiers – how to ID that which can be safely ephemeral from that which should be long-lived [Cerf]

Why not save everything, if storage is so cheap? [Kahn]

Because of the social questions/security [Cerf]

Which are a subset of the larger questions of how to manage intellectual property in the network age [Kahn]

Can be hard to make incremental changes; e.g., move from IPv4 to IPv6 [Kahn]

Unexpected uses and abuses: SPAM, advertising, virtually free email.  These sorts of things, esp. free email, can have a vastly different sort of use than originally envisioned due to the economics of the situation [Cerf]

What if the original basic assumptions could be changed?  Ideas are often constrained by technological and economic limitations, as theirs surely was.

The internet may not be the future of networking.  May have just been a first instantiation.

Issues of fundamental properties not cooperating; Interplanetary Internet
Try using TCP/IP to communicate with Mars.  Forty minute roundtrip at speed of light; what delay and fault tolerance?  The planets are moving in their orbit around the sun, and they are rotating which means the computer you are trying to communicate with may have just "disappeared."
Led to design of delay and disruption tolerant networks. [Cerf]

The only reason they could make progress on the internet back in the 70s was because "nobody cared or didn’t think it was important." [Kahn]

I’m not sure where the talk can be found on the internet, but it was sponsored by ACM and was part of the SIGCOMM 2005 Data Communications Conference.

Update: 23 Aug 05

From NPR’s Morning Edition 22 Aug 05: Computing Pioneers Discuss the State of the Net [Thanks, Jenny!]