[ih] Fwd: Some Berkeley Unix history - too many PHDs per packet

[Clem Cole]

This is UNIX history, but since the Internet's history and Unix history are
so intertwined, I'm going to risk the wrath of the IH moderators to try to
explain, as I was one of the folks who was at the table in those the times
and participated in my small way in both events: the birth of the Internet
and the spreading of the UNIX IP.

More details can be found in a paper I did a few years ago:
https://technique-societe.cnam.fr/colloque-international-unix-en-france-et-aux-etats-unis-innovation-diffusion-et-appropriation--945215.kjsp
[If you cannot find it and are interested send me email off list and I'll
forward it].

And ... if people want to continue this discussion -- please, please, move
it to the more appropriate COFF mailing list:
https://www.tuhs.org/cgi-bin/mailman/listinfo/coff - which I have CC'ed in
this reply.


On Fri, Mar 8, 2024 at 11:32 PM Greg Skinner via Internet-history <
internet-history at elists.isoc.org> wrote:

> Forwarded for Barbara
>
> > I will admit your response is confusing me.  My post only concerns what
> I think I remember as a  problem in getting BSD UNIX, in particular the
> source code. Nothing about getting something  we wanted to use on a
> hardware platform from one of the commercial vendors.  We needed the BSD
> source but got hung up.
>

Let me see if I can explain better ...

Assuming you were running BSD UNIX on a Vax, your team would have needed
two things:

   - an AT&T License for 32/V [Research Version 7 -- port to a Vax/780 at
   AT&T] and a
   - a license for BSD 3, later 4, then 4.1, *etc*., from the Regents of
   the University of CA.

The first license gave your team core a few rights from AT&T:

   1.  the right to run UNIX binaries on a single CPU (which was named in
   your license)
   2. the right to look at and modify the sources,
   3. the right the create derivative works from the AT&T IP, and
   4. the right to exchange your derivative works with others people that
   held a similar license from AT&T.

[AT&T had been forced to allow this access (license) to their IP under the
rules of the 1956 consent decree - see paper for more details, but
remember, as part of the consent decree allow it to have a legal monopoly
on the phone system, AT&T had to make its IP available to the US Gov --
which I'm guessing the crux of Barbara's question/observation].

For not-for-profits (University/Research), a small fee was allowed to be
charged (order of 1-2 hundred $s) to process the paperwork and copy the mag
tape. But their IP came without any warranty, and you had to hold AT&T
harmless if you used it. In those days, we referred to this as *the UNIX IP
was abandoned on your doorstep.*   BTW: This license allowed the research
sites to move AT&T derivative work (binaries) within their site freely.
Still, if you look at the license carefully, most had a restriction
(often/usually ignored at the universities) that the sources were supposed
to only be available on the original CPU named in their specific license.

Thus, if you were a University license, no fees were charged to run the
AT&T IP on other CPUs --> however, the licensees were not allowed to use it
for "commercial" users at the University [BTW: this clause was often
ignored, although a group of us at CMU hackers in the late 1970s famously
went on strike until the Unversity obtained at least one commercial
license].  The agreement was that a single CPU should be officially bound
for all commercial use for that institution.  I am aware that Case-Western
got a similar license soon after CMU did (their folks found out about
the CMU strike/license).  But I do not know if MIT, Standford, or UCB
officials came clean on that part and paid for a commercial license
(depending on the type of license, its cost was the order of $20K-25K for
the first CPU and an order of $7K-10K for each CPU afterward - each of
these "additional cpu' could also have the sources - but named in an
appendix for each license with AT&T).   I believe that some of the larger
state schools like Penn State, Rutgers, Purdue, and UW started to follow
that practice by the time Unix started to spread around each campus.

That said, a different license for UNIX-based IP could be granted by the
Regents of the University of CA and managed by its  'Industrial
Laison's Office" at UCB (the 'IOL' - the same folks that brought licenses
for tools like SPICE, SPLICE, MOTIS,* et al*). This license gave the holder
the right to examine and use the UCB's derivative works on anything as long
as you acknowledged that you got that from UCB and held the
Regents blameless [we often called this the 'dead-fish license' -- *you
could make a chip, make a computer, or even wrap dead-fish in it.*  But you
had to say you started with something from the Regents, but they were not
to be blamed for what you did with it].

The Regents were exercising rights 3 and 4 from AT&T. Thus, a team who
wanted to obtain the Berkeley Software Distribution for UNIX (*a.k.a*. BSD)
needed to demonstrate that they held the appropriate license from AT&T
[send a copy of the signature page from your license to the ILO] before UCB
would release the bits. They also had a small processing fee to the IOL in
the order of $1K.   [The original BSD is unnumbered, although most refer to
it today as 1BSD to differentiate it from later BSD releases for UNIX].

Before I go on, in those times, the standard way we operated was that you
needed to have a copy of someone else's signature page to share things. In
what would later become USENIX (truth here - I'm an ex-president of the
same), you could only get invited and come to a conference if you were
licensed from AT&T. That was not a big deal. We all knew each other.
FWIW: at different times in my career, I have had a hanging file in a
cabinet with a copy of the number of these pages from different folks, with
whom I would share mag tapes (remember this is pre-Internet, and many of
the folks using UNIX were not part of the ARPAnet).

However, the song has other verses that make this a little confusing.

If your team obtained a* commercial use license* from AT&T, they could
further obtain a *commercial redistribution license*.  This was initially
granted for the Research Seventh Edition. It was later rewritten (with the
business terms changing each time) for what would eventually be called
System III[1], and then the different System V releases.   The price of the
redistribution license for V7 was $150K, plus a sliding scale per CPU you
ran the AT&T IP, depending on the number of CPUs you needed. With this, the
single CPU for the source restriction was removed.

So ... if you had a redistribution license, you could also get a license
from the Regents, and as long as you obeyed their rules, you could sell a
copy of UNIX to run on any licensed target.  Traditionally, hardware is
part of the same purchase when purchased from a firm like DEC, IBM,
Masscomp,* etc*. However, separate SW licenses were sold via firms such as
Microsoft and Mt. Xinu. The purchaser of *a binary license* from one of
those firms did not have the right to do anything but use the AT&T
derivative work.  If your team had a binary licensee, you could not obtain
any of the BSD distributions until the so-called 'NET2" BSD release [and
I'm going to ignore the whole AT&T/BSDi/Regents case here as it is not
relevant to Barbara's question/comment].

So the question is, how did a DoD contractor, be it BBN, Ford Aerospace,
SRI, etc., originally get access to UNIX IP? Universities and traditional
research teams could get a research license.   Commercial firms like DEC
needed a commercial licensee. Folks with DoD contracts were in a hazy
area.    The original v5 commercial licensee was written for Rand, a DoD
contractor.   However, as discussed here in the IH mailing list and
elsewhere, some places like BBN had access to the core UNIX IP as part of
their DoD contracts. I believe Ford Aerospace was working with AT&T
together as part of another US Gov project - which is how UNIX got there
originally (Ford Aero could use it for that project, but not the folks at
Ford Motors, for instance].

The point is, if you access the *IP indirectly* such as that, then
your site probably did not have a negotiated license with a signature page
to send to someone.

@Barbara, I can not say for sure, but if this was either a PDP-11 or a VAX
and you wanted one of the eBSDs, I guess/suspect that maybe your team was
dealing with an indirect path to AT&T licensing -- your site license might
have come from a US Gov contract, not directly. So trying to get a BSD tape
directly from the IOL might have been more difficult without a signature
page.

So, rolling back to the original.  You get access to BSD sources, but you
had to demonstrate to the IOL folks in UCB's Cory Hall that you were
legally allowed access to the AT&T IP in source code form.  That
demonstration was traditionally fulfilled with a xerographic copy of the
signature page for your institution, which the IOL kept on file.   That
said, if you had legal access to the AT&T IP by indirect means, I do not
know how the IOL completed that check or what they needed to protect the
Regents.

Clem




1.] What would be called a System from a marketing standpoint was
originally developed as PWB 3.0.   This was the system a number of firms,
including my own, were discussing with AT&T at the famous meetings at
'Ricky's Hyatt' during the price (re)negotiations after the original V7
redistribution license.