[TUHS] Trade Secrets and Copyrights [was History of cal(1)]

[Clem Cole via TUHS]

On Sun, Sep 21, 2025 at 8:27 PM Theodore Ts'o <tytso at mit.edu> wrote:

> So inquiring minds want to know.  When was the first date that Unix
> was distributed outside of Bell Labs, and when was the first license
> which included the assertion that it was a trade secret?
>
Columbia had the first.   Lou Katz was >>external<< user 1 of Fourth
Edition [Doug wins the awards for internal use].  I'd have to ask Lou what
the date was, but I have always thought it was soon after Oct'73 SOSP
(which redated the July CACM paper).





>
> According to Wikipedia ("which is never wrong"),
>
>    In 1975, the first source license for UNIX was sold to Donald
>    B. Gillies at the University of Illinois Urbana–Champaign (UIUC)
>    Department of Computer Science.[27]
>
Interesting use of the words "sold" -- I wonder why .  I had thought the
first commercial license was for Rand.

>
> Note that 1975 is *after* the publication of the CACM paper.
>
> There are hints that the Unix source code was shared before the first
> license, which makes it even worse.
>
That is the first time I have ever heard that.   Do you have a source?

>
> But with a Trade Secret, it can only be protected if whenever you
> share it with someone outside of your organization, they must have
> signed an NDA, and the protection of the trade secret is if that
> person makes an unauthorized disclosure, (a) it's no longer protected,
> and (b) you can sue that person into oblivion.

I agree that your description of using an NDA is the legal
interpretation/practice.  But I do wonder if it was the same way in the
70s.  Remember also that AT&T is working under the 1956 consent decree that
requires, in return for having a monopoly on the phone business,
the corporation to "make its IP available to interested parties under
reasonable licensing terms."  The transistor is, of course, the best
example of that — and AT&T/WE hardly made the kind of money that RCA, GE,
TI, *et al *made creating transistors.



> ....
> The bottom line is that if Bell Labs had *ever* distributed the
> information to third parties without an NDA before trying to claim
> that the methods and concepts of Unix were a "trade secret", it simply
> was neer going to fly.

Well, I agree that trying to call it a trade secret after so many of us had
been taught the techniques in school seems silly.   And I never
understood how that claim could be made.   But I'm not sure that an NDA was
required when they started.  >>As I understand it<< the use of an NDAs
really started to become more the norm only in the late 1970s.

As I mentioned, we were warned about the AT&T copyright and its license,
which is why the CMU lawyers had us sign the sublicense if we took the OS
course.   But there was never anything associated with a NDA.  I did not
even become familar with the term until a few years after I graduated.



> And to the extent that Bell Labs had
> distributed Unix source code to MIT without the trade secret clause,
> it was also doomed --- which is probably why AT&T wasn't willing to
> tell third parties like say, Digital, that MIT had a valid Unix
> license,

Al Arms and the folks at AT&T patent and licensing would not verify anyone
having a license.  That was not an MIT-specific thing.  This is why we all
sent a copy of the signature page for the license to another site before
they would send you the code.   The original "UNIX User Group" and Mel and
Lou set up (which became USENIX) was invitation-only.  That's why Ken put
Mel's address in the V6 kit and told you to contact Mel.   To be invited,
you had to send Mel a copy of your signature page.



>
> The bottom line is that trying to use Trade Secre was unbelievably
> stupid, and suggests that either (a) Bell Labs and AT&T Lawyers were
> incompetent, or (b) they weren't aware of prior dissemination of the
> information that they were trying to claim were trade secret.  I
> suspect (b) is much more likely.  Perhaps they never read the Unix
> CACM paper.  :-)
>
Well, that's stronger than I would state.   The AT&T legal team was aware
of the disclosures and had been required to license and disclose them via
the 1956 consent decree.   As was explained to me, that was the basis of
their claim.  It was there IP.  They owned it (and still did), but at the
time, the law required them to share it, which was a bit of a catch-22 for
them.  Thus, the open question became "could they still call the IP a trade
secret after having shared it?"  The court said no.  Anything they invented
after Judge Green did not have to be shared, but before then, which UNIX
certainly was.

So, the interesting question is between the time of the 1956 decree and
Judge Green: could AT&T realistically have had any trade secrets if it were
required to license its IP?


> It's also not surprising that hacker community got it wrong, because
> it was such an insane and stupid legal tactic!
>
Again, as I said, we have spent all of the time being concerned about
protecting AT&T's license and the copyright to their example implementation
of that IP.