UNIX trademark registration

[Henry Spencer]

> I do not fully understand how AT&T is able to hold nearly all UN*X systems
> as a trade secret.  Is there some kind of a limit on how wide spread
> their 'trade secret' can be extended?  Is there some method which
> a trade secret becomes so widely known that it is no longer secret?
> (much like a trademark can become public domain via too much generic use)

A trade secret becomes unsecret only if the secrecy is broken.  That is,
if the secret becomes widely known to people who have *not* formally agreed
to keep it quiet.  This happens if AT&T gets sloppy about protecting it
(in which case they aren't treating it as a secret, and their protection
vanishes), or if somebody who has formally committed himself to keeping
it quiet (i.e., by signing a Unix licence) spills the beans.  In the latter
case, the man who spills a secret he has agreed to keep secret is going to
get sued for his shirt.  So is the man he spills it to, if said recipient
should have known it was secret.  If AT&T catches it quickly -- and they
are indeed on the alert for such breaches of secrecy -- they may well be
able to contain the leak.  If they can't, the secret isn't a secret any more.

However, so long as the secret is protected properly, by things like
non-disclosure agreements, there is no limit on how widely it can be used
while still remaining legally a trade secret.

> How is AT&T able to claim trade secret over something like 4.2BSD?

Easy.  It's derived from Bell code.  And Berkeley, like all of us, has
signed a non-disclosure agreement that requires keeping the Bell code
secret.  So long as there is one byte of Bell code in x.yBSD, it is still
covered, and if Berkeley wants to distribute it, they have to be careful
to do so only to people who are authorized to see the Bell parts.

> If you diff 4.2BSD source with V.2 source, you will find that
> a great deal of the code is not the same.  Also BSD got major
> funding via DARPA, who is in turn funded by US tax payers.
> Is there some point where the long arm of trade secret breaks?
> Does who did the work and paid for it impact the status of another
> person's trade secret claim?  If you start out with a trade secret
> code and hack it up one side and down the other while the original
> code goes off in another direction, can you claim your own trade secret?

It doesn't matter that some of it is different; what matters is that some
of it is still the same.  Another thing that matters is that the Berkeley
work is based on knowledge obtained by studying the secret material,
although *this* is a very slippery area.

Who else contributed to the development of x.yBSD is irrelevant; DARPA
has no power to declare AT&T's trade-secret protection invalid.  They
could impose *additional* restrictions if they wanted to and could clear
it with their superior authorities, but they can't arbitrarily disregard
software agreements they signed to get Unix.

Similarly, you can claim your own trade secret protection for mods you
make to x.yBSD or System N, but this does not eliminate the trade-secret
protection AT&T has made you agree to as a condition of giving you the
stuff.  If you rewrote the stuff from scratch, in a distinctly different
way (so you weren't just reconstructing the Bell stuff from memory),
*then* AT&T has no claim on you.  This is what the Mark Williams folks
did.  But they were careful:  the innards of their system are not at all
similar to the Unix kernel, and they did not have access to Unix source
while they were developing it.  (Having access to the source doesn't
prove that you looked at it, but it makes it a whole lot harder to prove
that you *didn't* if AT&T sues you for copying their secret.)

[Warning:  I am not a lawyer.  For heaven's sake, consult a specialist
before doing anything rash!  I disclaim all responsibility...]
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry