Information Technology Reference
In-Depth Information
By early 1968, Roberts had nearly completed a detailed design for the net-
work. There were three basic principles (see
Fig. 10.13
). The first was that the
IMPs should operate as a communication system whose essential task was to
transfer packets from anywhere to anywhere else on the network. The IMPs
would also take care of route selection and acknowledgment of receipt for the
packets. The second requirement was that the network must have a very small
message delay time. From his experience with time-sharing systems, Roberts
decided that the average transit time through the network must be less than
half a second. The third principle was that the IMP system should function
independently of whether a host computer had crashed. Network reliability
should depend on the IMPs and not the host computers. In addition, the NPL
team from the United Kingdom had convinced Roberts that he should specify
the use of much faster links than he had specified in his original proposal. By
July 1968, Roberts sent out a formal request for quotation to 140 companies
giving them information about the project's requirements and inviting them
to bid on building the IMPs. One early setback was that two of the major com-
puter companies, IBM and Control Data Corporation (CDC), declined to bid
and said that “the network could never be built because there existed no com-
puters small enough to make it cost-effective.”
17
Fortunately, Roberts eventu-
ally received more than a dozen bids. In December 1968, ARPA announced that
it was awarding the contract to build the IMPs to a small consulting firm in
Cambridge, Massachusetts, called Bolt, Beranek and Newman, or BBN.
BBN started life as a small consultancy group advising on acoustics in
1948. Richard Bolt and Leo Beranek taught on the MIT faculty, and both were
experts on acoustics and, in particular, the acoustics of buildings. Robert
Newman, who had been a student of Bolt's and was an architect, joined the
consultancy a year later, and BBN was born. The business grew rapidly, and
besides its profitable work on buildings, BBN developed unique expertise in
the analysis of audiotapes. The company assisted in the analysis of the film
of the shooting of President John F. Kennedy and of the shooting deaths at
Kent State University. Most famously, in the wake of the Watergate scandal,
the White House and the special prosecutor's office called in BBN to examine
the 18½-minute gap in the Nixon White House tapes. Dick Bolt headed the
investigating committee, which concluded that the erasure was deliberate.
By the time that BBN recruited Licklider in 1957, the company had a well-
established hiring philosophy summarized by Beranek: “I had the policy that
every person we hired had to be better than the previous people.”
18
BBN was
also well known for its policy of hiring MIT dropouts. The rationale was that if
the person had managed to get into MIT, he or she must be smart, and, if the
person had subsequently dropped out, all this meant was that he or she could
be hired more cheaply than someone who had graduated. Because of the com-
pany's recruiting policies and because of the absence of any academic tenure
process and any teaching commitments, BBN became a very attractive place
for researchers to work. In this way, BBN acquired the informal reputation of
being the “third university” in Cambridge, along with Harvard and MIT.
When Licklider joined the company, he convinced Beranek that they should
buy him a computer. Beranek later said, “I decided that it was worth the risk to
spend $25,000 on an unknown machine for an unknown purpose.”
19
The gamble
Fig. 10.13. Diagrams of the host nodes
and the IMP network: (a) the host nodes;
(b) the message processors that allow
interconnections through the network;
and (c) each message processor can be
the messenger between two other mes-
sage processors.
Search WWH ::
Custom Search