“That allowed some people to say, ‘Okay, it
might be helpful for nuclear survivability’, and
the person listening might be a senior person
who says ‘We have to spend money on that’. If
you ask, ‘If we had been trying to build a nuclear-
survivable system, what would we have done?’,
then you look at the problems we focused on,
there are big differences. But you can peel that
apart, and ask, ‘Are we using this technology for
nuclear-survivable networks today?”
By coincidence, Crocker went to the same
high school as Vint Cerf, who briefly worked on
the Apollo program before moving to UCLA and
the laboratory of Professor Leonard Kleinrock,
where the first two nodes of the ARPANET were
connected. It’s the same school as Jon Postel,
who was involved in the development of the
internet domain system at UCLA and directed
Cerf and Robert Kahn to develop what’s now
known as Transmission Control Protocol, TCP/IP.ONE MILLION DOLLARS
On his desk in ARPA HQ in Virginia, the agency’s
director of information processing techniques,
Bob Taylor, had three computer terminals, one
each for the separate computer systems ARPA
was funding. One was in Santa Monica at the System Development
Corporation, one in Berkeley at the University of California,
and one at MIT. “For each of these three terminals, I had three
different sets of user commands,” Taylor was quoted as saying
in the New York Times. “So, if I was talking online with someone
at SDC, and I wanted to talk to someone I knew at Berkeley, or
MIT, about this, I had to get up from the SDC terminal, go over
and log into the other terminal and get in touch with them. I said,
‘It’s obvious what to do—there ought to be one terminal that goes
anywhere you want to go’. That idea is the ARPANET”.Taylor managed to get a million dollars
redirected from a ballistic missile budget and
hired Larry Roberts as program manager. At a
design session in April 1967, Roberts proposed
that mainframes could connect to one another
directly. This would take computing capacity
away from the mainframes and dedicate it to
network processing, so a counter-proposal was
made by Wesley Clark, the man often credited
with designing the first modern personal
computer, who suggested mini-computers
should act as interfaces instead. The ARPANET
plan was modified, and the mini-computers
named Interface Message Processors (IMPs).
This idea was taken to Tennessee for the
Symposium on Operating Systems Principles
in October 1967, after which Davies’ packet-
switching ideas were absorbed into the project.
Roberts later said ARPANET and other packet
switching networks from the 1970s were similar
in nearly all respects to Davies’ 1965 design.
The specification for the IMPs was finalized
the following year, and a request was sent out to
140 computer companies asking them to submit
quotes for building the network. So outlandish
was the idea that only 12 replied, with four of
those considered as ‘top-rank’ contractors. In the end, only two
quotations were considered and the job was awarded to a team
from Bolt, Beranek, and Newman (BBN), led by Frank Heart and
Robert Kahn, who’d completed a Ph.D in electrical engineering at
Princeton in 1964. He worked at BBN, then moved over to ARPA’s
Information Processing Techniques Office where he implemented
the SATNET early satellite network.
That lab brought together two previously distinct disciplines;
computer science, of which there were many practitioners
from the universities, and communications. “That was me,”
This building in
Arlington, Virginia,
was DARPA HQ until
2009, when it moved
its operation closer
to the Pentagon.UCLA’s Interface Message
Processor (IMP) is pictured in
a storage closet, where it had
been hidden for over 20 years.APR 2022 MAXIMU MPC 41
© WIKIMEDIA/
COOLCAESAR, WIKIMEDIA/ STEVE JURVETSON