[ih] Historical fiction

Craig Milo Rogers rogers at isi.edu
Fri May 11 09:29:34 PDT 2012


On 12.05.10, Alex McKenzie wrote:
> Every site connected to the early ARPAnet was the center of something.
> 
> IMP 1: UCLA was the place where Kleinrock worked - applying 
> queueing theory to data communications.  Because of this he had an ARPA 
> contract to operate the Network Measurement Center which used facilities built into the IMPs (packet tracing, periodic measurements of queue 
> lengths and waiting times, traffic statistics, traffic generators, etc) to run experiments with the performance/behavior of the "subnetwork".
>     UCLA was also home of the Campus Computer Network and its IBM 360/91, funded in large part by ARPA.  This was the largest commercial number-cruncher planned to be connected to the network, and was therefore an important 
> number-crunching resource to be shared by the ARPA research community.

	UCLA was also the home of the Center for Computer-based Behavioral
Studies (CCBS), a research unit of the UCLA Psychology Department that was
essentially an SDC (Systems Development Corp. in Santa Monica) spinoff (SDC
was itself a RAND spinoff), funded by ARPA.  Initially located in the basement
of the Business School building, Bunche Hall, CCBS soon moved to a
purpose-built area on the third floor of Middle Franz Hall.  CCBS had the
first PDP-10 (KA10) system assembled with a full complement of 256K 36-bit
words of core memory.  This was bigger than the CTSS and ITS systems at MIT, a
bit smaller than the 384K 36-bit words on MULTICS, but only a quarter of the
main memory capacity of CCN's IBM 360/91.  CCBS was chartered to offer general
purpose timesharing services, interactive statistical analysis tools, and
realtime, online studies of collaboration and competition (wargaming, in
piecemeal), both in-house and to military and academic customers over the
ARPANET.

> IMP 3: UCSB had an IBM 360/75 which was used to run a time-sharing system, 
> which included special CRT terminals designed for the UCSB system (at 
> this time almost every ARPAnet site except UCSB and SRI used hard-copy 
> terminals).  The UCSB  360/75 was a major resourch intended to be 
> shared.

	CCBS also built special CRT terminals, based on the Tektronix storage
tube technology, with light pens and other special features intended to
compete with the PLATO system.  I can't remember if they were fully
operational in 1970, but I'm pretty sure they were in use by 1971.  In a
sense, this was a continuation of the group's work at SDC, which had CRT
terminals (fed by a drum storage unit) (and with light pens for input) on the
IBM AN/FSQ-32 all-transistorized timesharing system (funded by initially by
SAC to replace the vacuum-tube AN/FSQ-7 computers, taken over by ARPA to study
timesharing), which I used in 1968.

	There was a dedicated computer-computer link between SDC and the TX-2
at MIT.  This was a significant pathfinder effort prior to the ARPAnet.  I saw
it in 1968, although I didn't get to poke at it.  Underlying this West Coast
activity was SAGE, the Semi-Automatic Ground Environment, a military defense
program, which was the true pioneer in developing timesharing, computer
communications, CRT-based interfaces and large-scale computing.

	You might ask, why wasn't SDC put on the ARPANET?  It is my
understanding that even after the group that became CCBS moved from SDC to
UCLA, it was intended to tie SDC into the ARPAnet fairly quickly.  However, a
dispute arose between SDC and the government; I recall being told that SDC
attempted to retroactively raise its overhead rates on its research contracts,
leading to a drastic reduction in SDC's ARPA funding and the cancellation of
SDC's ARPAnet connection.

					Craig Milo Rogers



More information about the Internet-history mailing list