These are some notes made in 2005 on a paper by Dave Pehrson in about 1970 which includes much information that I had feared had been lost.

The paper could be called “The LAN (or SAN) before Ethernet”. (Or the pre byte LAN)

There are several points that might confuse the modern reader. Some hermeneutics follow:

A “transmission line” is a line where there are more than one bits in the metal conductor at a time. There were few if any transmission lines in the Octopus system. A teletype line a thousand miles long is a transmission line. Ethernet is a transmission line and required very specialized hardware. I suspect that TMDS had transmission lines between controller and monitors. The CRAY I and II had transmission lines between the CPU and mass store. The Livermore network began to use transmission lines soon after this note was written. They discovered that two packets could pass each other on the same physical conductor when the line crossed the campus (about one mile).

The PDP-6 or a worker could produce a raster image and transmit it to the TMDS, One of the 16 active images being maintained by the TMDS could be changed during any one frame time. It was not for movies, but for frequent stills. When I saw it running a system status image appeared on otherwise unused monitors.

I was glad to see the information in table 1. I had forgotten most of that. One figure of merit that has perhaps been lost is accesses per unit time. It may be greater than the inverse of access time. The table notes that the photostore access is overlapped. The Librascope disk file was overlapped as well in that several queued accesses could be served in a single revolution. That was the advantage of the fixed head over the movable head 3330. Of course what I say is only theory. There may have been barriers to achieving this such as

Here is a detailed architectural description of the Livermore Octopus. They are much evolved from the ideas when I left.