Subject: Re: termless copyright and patents
From: <stephen@xemacs.org>
Date: Tue, 26 Sep 2006 20:34:28 +0900

Thomas Lord writes:

 > FC is about getting works into the public domain during "our
 > lifetime" but it does pretty much nothing to create incentive for
 > author's.  At one end of a spectrum, FC licensing is probably just
 > fine (but completely unimportant) for, say, the entire catalog from
 > O'Reilley.  At the other end, if you approach a big publisher with
 > your great american novel, FC is just a big strike against you.

And a limited patent wouldn't be?

 > Limited patents, on the other hand, are about using markets to pay
 > inventors for putting something in the public domain just when its
 > needed.

That's the way it worked out in your scenario, but I don't see why it
would necessarily work out that way---not even occasionally.  See below.

 > It's funny you should say that because I've been trying to decide
 > whether or not covenants made by IBM, MSFT, et al. are entirely
 > altruistic

I would dismiss that hypothesis out of hand.

 > or whether they aren't being made precisely to stave off a court
 > case in which defensive patent portfolios lose all value.

What scenario do you have in mind?

 > For example, IBM may discover that, in shipping Linux systems, they
 > have practiced many of their own patents.  Well, per RMS, those
 > patents would all now be publicly licensed for any purpose
 > whatsoever.

First, that's not true; it would only be licensed for GPLed software.
I'm not sure how good an approximation that is to "any purpose
whatsoever".

Second, even if it's a good approximation, per Larry Rosen's book
AIUI, it's still not true; the broad implicit license RMS claims is an
hypothesis with no visible support in case law.  Eben Moglen is a big
lawyer, but IBM beat the whole Justice Department, more than once.
I'll put my money on IBM if the FSF sues.

 > And, for example, MSFT would be nuts to try to shut down something
 > like Samba (hypothesizing it must violate some MSFT patent or
 > other) but unless they explain *why* Samba isn't being sued then
 > haven't they, too, implicitly given unrestricted public license?

No.  It's at the option of the patent holder who to sue or not, and he
can change his mind at any time, even a full decade later.  Did you
miss the whole LZW episode?

 > And so these firms (and Mr. Rosen) are now in the interesting position
 > of trying to define "open source" in way that doesn't quite accord
 > with either OSI or FSF definitions.   Huh.

Larry has denied that several times.  Why don't you believe him?  The
point being that the software as a whole being OSI "open source"/FSF
"free" is something he's just plain not interested in discussing,
AIUI.  The important thing is that *third parties* must license
*their* portion under such terms, and the result will distributable
under the combined conditions.

There is a question in my mind, too, whether you could usefully
license software practicing the patent under the GPL, version 2 or
draft version 3---downstream might not be able to redistribute it.
(This would also imply that the software would have to be wholly
original---you wouldn't be able to add code practicing the claim to
software licensed to you under the GPL.)  But you certainly could
license it under an academic license, and downstream would be able to
redistribute under the same terms.

(I note that Larry has denied the restriction against the GPL as well,
saying that he believes the covenant is compatible with the GPL.)

 > As for characterizing limited patents as being about a "lifestyle
 > business," I'm really not sure what you mean (other than something
 > derogatory).

"Charitable."  I don't consider "charitable" to be a derogatory term.

 > Why would, say, the research dept. at HP not want to use limited
 > patents?

Because anything they can do with a limited patent they can do with a
conventional patent and a covenant or a license, at a time convenient
to them.

 > Would it be cheaper to buy them outright?   Well, you can find my
 > opinion in the scenario I described.

Scenarios are even cheaper than limited patent licenses.

I can give you an alternative scenario: MSFT offers $475K for an
exclusive license on a conventional patent to poke IBM, Sun, and you
in the eye simultaneously.  The patent goes into the public domain in
20 years....

Or IBM says "we'll pay your $165,000 asking price based on amortizing
over 5 years, but in return we want a veto over the public domain
offer before 5 years."  Alice says, "no," and they come back with,
"Cool.  We expect that option to be exercised in 2 years.  Will
$75,000 do?"

Why is your scenario more plausible than either of mine?  Where do the
numbers you sling with such abandon come from?  (Mine come from yours,
with a little interpolation in the second scenario.)