[ih] Hourglass model question

[Dave Crocker]

On 7/5/2019 10:09 PM, Vint Cerf wrote:
> John, i had the same impression - that there was demand for new 
> computing equipment and ARPA wanted the research groups to be able to 
> share resources as well as sharing code and research results freely.


I can't speak to the motivations at the base of the Arpanet's origin 
story, but certainly this desire (possibly rising to the level of goal) 
was present by the early 70s.

And thanks so much for giving me the opportunity to recite the proof, 
which will be especially close to Steve's and Vint's heart...

The UCLA project used an XDS Sigma7, running a locally-developed 
time-sharing system, which I believe was formally called the Sigma 
Executive system.  One of my earliest assignments was to develop 
documentation for it.  That process was how I learned about operating 
systems, including comparisons with Tenex.

Eventually and inevitably it needed more main memory and a source of 
some second-hand memory was located.  Arpa was asked for the additional 
funding to buy it, which I believe was relatively modest.  Instead Arpa 
noted that our o/s was unique to UCLA, which meant that it was a dead 
end, but they noted that there was an entire network of resources 
available through the Arpanet.

Even better was that there was the recent availability of PDP-11 based 
network access host systems (Illinois' ANTS and UCSB's ELF).  They told 
us to get a PDP-11 and switch over to using one of those systems and 
relying on network-based computing.

Both ANTS and ELF first versions were done as hacks, but were 
immediately successful.  So version 2 of each was funded.  Both Version 
II efforts were, shall we say, problematic, including highly delayed.  I 
was on an oversight committee that was eventually formed for the ANTS-II 
project and we were in Illinois the day the official analysis of the 
Nixon Tape's 18 1/2 minute gap was announced publicly, citing the 
analysis as having been done by an obscure acoustics research company in 
Cambridge MA, namely BBN.  Jerry Burchfiel, of BBN, was also on our 
committee and quietly told us some stories about that stealth project.

We did take delivery on ANTS-II but it wasn't very usable.  However 
there was a new PDP-11 based operating system that had emerged, called 
Unix, and the folks at Illinois quickly developed an NCP for it.  At 
UCLA we immediately switched over to using it.

The UCLA o/s effort had developed an elaborate naming model for the 
systems and its components.  I'll let Steve and Vint decide whether to 
go into that detail, but if they do, Steve has to tell about our 
father's contribution.

Anyhow the acronym for the Sigma Executive was the SEX system.  Besides 
doing basic documentation for that I was tasked with compiling 
documentation for our use of network resources.  I'm not sure whether I 
named that compendium or someone else did, but it was the Network Use 
Technical Series.  That is, NUTS Notes.  Steve once commented that it 
was the only document that had joint authorship by him, me, Vint and Jon 
Postel (plus some others, of course.)

So Arpanet made us get rid of SEX and use Unix.  I suppose it was 
inevitable that the first superuser password for our new system was... 
eunuchs.

d/

-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net