
Larry Roberts, ARPA; 3. Steve Carr, UCLA; 4. Jeff Rulifson, UTAH; 5. Ron Stoughton,
UCSB; 6 Steve Crocker, UCLA.
7. Robert Kahn, “Scanning the Issue: Special Issue on Packet Communication Networks,”
Proceedings of the IEEE, Vol. 66, No. 11, p. 1303.
8. “An Interview with Leonard Kleinrock,” conducted by Judy O’Neill, 3 April, 1990.
Charles Babbage Institute, The Center for the History of Information Processing,
University of Minnesota, Minneapolis.
9. Ibid.
10. Ibid.
11. From an Interview with Robert E. Kahn, conducted by Judy O’Neill, on April 24,
1990, Reston Virginia, Charles Babbage Institute, Center for the History of Information
Processing, pp. 18-21. Kahn Said:
“When I got there [DARPA] there was money budgeted for a packet radio program, and
I undertook to make it happen. The skids were all greased for that. Part way through the
first year of the program it became clear to me that we were going to have to have a plan
for getting computer resources on the net. In 1973, mainframe computers were multi-
million dollar machines that required air-conditioned computer centers. You weren’t
going to connect them to a mobile, portable packet radio unit and carry it around.”
“So my first question was ‘How am I going to link this packet radio system to any
computational resources of interest?’ [Kahn had just succeeded in solving that question
with the ARPANet at the ICCC72 show]. Well, my answer was, ‘Let’s link it to the
ARPANet.’ Except that these were two radically different networks in many ways. I
mean, all the details were different. I don’t mean conceptually they were different. They
were sort of the same genre. Just like, say Chinese and Americans are of the same genre
except one speaks Chinese and one speaks English, one lives on one side of the world,
one lives on the other side, they go to sleep during your daytime, etc. The details of the
two networks were rather different. The ARPANet ran at 50 kilobits per second and the
packet radio system ran at 100 or 400 kilobits per second One had thousand bit uncoded
packets; the other had two thousand bit packets which could be coded. The ARPANet
assumed that once you sent something it was delivered with a hundred percent reliability.
The other assumed that much of the time you would never get anything through even
though the system was working. The protocols that were designed for the ARPANet
wouldn’t work over the packet radio net because when a packet entered the packet radio
net, the only thing the ARPANet would have told it was where it came from but not
where it was going. So the packet radio net had no further information to know where to
route it. If a packet got lost along the way, the ARPANet hosts would come to a halt.
Well, in a radio net you can get interference and so some loss is natural So we really had
to rethink literally the whole issue of host transport protocols. Vint Cerf and I jointly
came up with the TCP/IP concept as a new transport mechanism as part of an architecture
for internetworking. DARPA then gave a contract to Vint at Stanford to actually
implement the TCP/IP concept - along with small efforts at BBN and at University
College London. Vint had the lead for developing the specification.”
12. Katie Hafner and Matthew Lyon, When Wizards Stay Up Late: The Origins of the
Page 91