Again I agree with much of what you said. But there
are a couple of
things your are forgetting....
On Wed, Feb 7, 2018 at 10:13 AM, Theodore Ts'o <tytso(a)mit.edu> wrote:
Well, MIT actually had multiple AT&T Licenses which did *not* have the
mental contamination clause.
I fear, you are
falling
into an
error of thinking about UNIX as the *source cod*
*e from Murray Hill* as opposed the* intellectual property *->* i.e.* an
*implementation* *vs,* *the ideas* of how to build the a computing
system. The ideas was
the point of the AT&T case
with BSD (as a number of people including some on this list have said --
AT&T was talking about trade secrets - which are ideas). Note I failed to
understand this difference myself at the time, most of us did also I fear
because as you said -- we all just wanted Unix for a PC based platform that
we own for ourselves. However, I believe that understanding that
difference is why the AT&T *vs*. BSDi/UCB case is so important to
history,
and Linux's history in particular.
What
doomed
AT&T was they were required by law to release the IP (1956 consent
decree)
to any interested parties for a fair and reasonable license, but they
had to continue to make their IP available to the academic community at no
charge [I site the paper from one the law reviews that is really detailed,
so I'm summaries the results. Send me a message offline if you want to
know more]. The fact is this worked out fine for technologies like the
transistor. People licensed it, AT&T never really made the kind of money,
TI, Fairchild, *etc* did, but it was AT&T IP that used. Folks took out a
license and the wider electronics industry was born from that invention.
However, later
AT&T's own personal start to publish papers and books about the
UNIX
IP
in the open literature.
It is these two acts together that the court said, meant that AT&T
could not longer claim trade secret - they licensed it AND they told people
about it. They 'taught' the world - mentally contaminated us all - with
their ideas. Linus for instance says he had Maury Bach's book when he was
building Linux and I have been lead to understand that Maury's book was
used as evidence in the case ( as was the original 1974 paper). The point
is once the IP was generally known, it was no longer a secret and AT&T
could not claim same. It also why the famous 6 files had to be removed
from the implementation that BSD had. Those were bases on the copyright
ownership and AT&T was allowed to claim copyright (as many people like me,
had expected them). They just could not claim copyright on the parts that
UCB had developed - even if they were the same ideas (IP) and AT&Ts.
So, I'm trying to be careful... this discussion was how did the Linux
implementation of the UNIX IP displace the BSD implementation of the same
IP? I believe that it is a classic example of a Christensen disruption...
a new market (PC based UNIX) was created nd people wanted something to fill
it. Linux was there to file it and did a great job and AT&T stopped
any alternative (BSD on same) long enough, that it did not matter - Linux
had filled the void well. The sustaining
technology
(BSD) stays with those that could afford it on their platforms (Sun,
DEC, *etc*...)
Stepping back a bit ... an interesting question is what about Sun's 386
attempts? The original RoadRunner was a pretty slick machine. But here is
where I think they got stuck the same way DEC and other did. They made
their money on the HW but it was the SW that drove the sale. So they
wanted to SPARC et al. So they kill RR. By, the time of they retrench and
Solaris/386 was released for "commodity HW", again - too expensive for the
low end and they did not want the Intel chips on the servers to compete
with their own HW.
Clem
As a minor footnote, I've always wonder about the SCO case being anything
but a delay tactic against Linux. Give the courts results from the AT&T*
vs*. BSDi case, I'm personally surprised they got as far as they did. I
really would expected to be tossed out at the start (hut Im not a lawyer of
course). Clearly, the courts had declared that ideas were free and clear
for anyone to use. So how SCO could have claimed anything against anyone,
I find hard to fathom. Again the courts got it
right
, but it took a loot of time and effort for then to decide something that
the another court had already decided.
Clem's entire post is too good not to quote in its entirety.
But it raises a question I've had for some time: what about clones like
COHERENT? COHERENT was kind of primitive by BSD (and even Linux) standards,
but had a lot of functionality and was pretty cheap (4 floppies for $99 or
so) and ran on more or less standard PC hardware; I'm surprised more people
weren't running it at home. Also, it seems that they'd weathered the storm
of an AT&T legal challenge managed to stay afloat, as related in Dennis
Ritchie's famous telling of it:
I wonder why the unchallenged existence of Unix clones like COHERENT and
Minix wasn't enough to kill the trade secret argument before it even got
out of the gate.
Clem, is your paper online somewhere?
- Dan C.