monday, june 09, 2008
How UCSB got to be an ARPAnet node
A while ago, somebody told me that UCSB had been the third ARPAnet node. This is pretty cool, but I didn’t know how or why it happened. The first two nodes were at UCLA (with Leonard Kleinrock) and the Stanford Research Institute (with Douglas Engelbart), and my school doesn’t seem to logically follow.
Then a month ago, a person who helped UCSB connect to the ARPAnet, Larry Green, came to my college to tell us all about it. He was an engineer at UCSB from 1965 to 1977, and he led the team that designed the interface between the university’s IBM 360 mainframe and its Internet Message Processor, a piece of hardware that connected to the other nodes of the network. I found an old description of a podcast with Larry, but the images and the link to the podcast are broken, so here are some pictures from his slides (which he kindly emailed to the people who attended) and my memory of what he said. I liked his friendly presentation; he balanced technical detail with human stories.
First Larry talked about early computers and the beginning of the ARPAnet, including a look at the log of the first successful net connection (the third entry sounds like the student closing up in order to go home). This general history is covered in a recent Vanity Fair article and the National Science Foundation’s “Birth of the Internet” special report, but neither of those publications mention UCSB more than in passing.
Next Larry explained the important Interface Message Processors, which were these big heavy boxes that functioned something like routers, handling the information sent between the different kinds of computers at different campuses. They had lifting rings on the top for potential helicopter transport or lowering into ships — this was still a Pentagon-related project.
It came in handy that the first ones were ruggedized like that because UCSB’s IMP traveled from Los Angeles International Airport to Santa Barbara by pickup truck. UCSB had invested in this hardware because it already owned a mainframe computer and there were some interesting people messing around with it. Larry and his team worked to make the big IBM 360 computer and the IMP communicate with each other, and that allowed UCSB to connect to the ARPAnet in 1969. (Larry explained that the diagram of the first four nodes looks hand-drawn because it was made using a light pen.) His team ended up selling a dozen IMP interfaces to other organizations that owned IBM 360s, so that they could connect to the ARPAnet too:
The protocol that connected the IMPs to each other was called BBN 1822, and Larry explained this diagram of it:
Larry also showed us this cartoon ad that ran in Datamation and said that yes, most of the people working on ARPAnet at the time were middle-aged white men:
UCSB was involved with all of this partly because of Glen Culler, a math professor who was inspired by J. C. R. Licklider and saw computers and networks as a way to help expand human minds. He developed the Culler-Fried On Line System, which ran on 65 classroom workstations connected to the IBM 360, all designed for teaching math. Quoting from Larry’s slides, the system’s “technologies of interactive graphics and on-line time sharing were exciting developments and the reason that UCSB was selected as an original ARPAnet site.” UCLA and SRI had nice important research, but I like that we had a very early online education system.
The IBM 360, the IMP, and the workstations were all located in North Hall on campus, and one of our professors said we should go over there and poke around in the basement to see if we can find any forgotten remains.
