Subject: Re: termless copyright and patents
From: <>
Date: Fri, 6 Oct 2006 14:17:26 +0900

Thomas Lord writes:

 > A library like your proposed "libsection11" has many uses which
 > do not arise from the exercise of GPL rights of the program which
 > Alice conveyed.

Uh, like what do you have in mind?

 > You would like "essential patent claims in" a work to include the
 > cutting and pasting of code into "libsection11" -- but there is no
 > language in the license that makes that obviously so.   

Excuse me?  Absent patent claims, it is your right under the GPL to
cut and paste code into libsection11.  Section 11 covenants that
Alice, who is conveying the program under the GPL, will not assert any
of the essential claims in that code due to exercising your rights
under the GPL.  Please complete the syllogism.

 > Alice is only required, by GPLv3 draft 2, to not assert against
 > infringements arising from the program she conveyed.

Right.  But AFAICS, per section 10, those downstream users have
received a license *from Alice*, specifically including that covenant.
Under section 11, then, Alice has covenanted not to assert the
*claims* in Alice's Program if they exercise their rights under the
GPL in *any* program with a chain of title going back to a Program
copy conveyed by Alice.

Probably the wording of section 11 should be clarified to cover
"derivatives practicing the essential claims of the conveyed Program."

 > First, "could" is misleading.  IBM is already promising not to
 > assert a set of listed patents against any open source code, at
 > all.  It's a hop-skip-and-a-jump from there to a "suggested claims"
 > covenant.

No, it's not, for all the "frictional" reasons that patents suck.
(N.B.  An "essential claims" covenant is equally far from an explicit
list of patents.)

 > And this brings us back solidly to reality and *why* so many large
 > companies are starting to make covenants on their own without
 > waiting for GPLv3.
 > I think it is *precisely* to protect value for their stakeholders.

Sure.  But **they** explicitly choose what they want to license, and
the burden is on users to determine whether their other IP might be
infringed.  Under the implicit covenant, they implicitly choose, and
the burden is on them to determine what unintended IP might be at
risk.  Big difference there.

 > I am inclined to think that GPLv2 actually carries a "suggested
 > claims covenant" implicitly.

Yeah, just like the 17-year-old boy is inclined to think his date
wants to be groped.  I suspect that patent-holders tend to think of
open source hackers as a law-abiding lot who would never intentionally
infringe somebody else's IP, and are shocked, *shocked*, I tell you,
at such broad claims of implicit licensing.<wink>

 > Distribution under these licenses is, in essence, a declaration that
 > the distributor expects people to use the source in all the ordinary
 > open source ways.

Sure.  But the "ordinary open source ways" don't include infringing
others' IP.  For example, you cannot mix SSLeay-licensed code with GPL
code, even though both are open source and you might think that mixing
open source code with open source code is what open source licenses
are all about, and even though pretty clearly eay was just being a
butthead about it to make a political point.  There is no difference
in principle between doing a patent search for the algorithms embodied
in a program and looking in the header comments to crosscheck the
compatibility of the copyright licenses.

Of course, there's a huge difference in expense.