MPL section 2.2, and patent grants on derivative works

Fri Nov 22 17:20:06 UTC 2002

Brian,

Most of the free / open source software licenses out there appear to be
pretty much equivalent vis-à-vis patent grants, in that those that are
silent, GPL, MIT, Apache, are likely granting implicitly what the others,
Mozilla, IBMPL, grant expressly.  By this I mean that patent law, under the
doctrine of implied license, gives to each distributee of a patented article
a license from the distributor to practice the patent claims owned by the
distributor covered by the patented article and any "reasonably contemplated
uses" of the patented article.  

To quote the Fed Cir, "Generally, when a seller sells a product without
restriction, it in effect promises the purchaser that in exchange for the
price paid, it will not interfere with the purchaser's full enjoyment of the
product purchased. The buyer has an implied license under any patents of the
seller that dominate the product or any uses of the product to which the
parties might reasonably contemplate the product will be put."
Hewlett-Packard Co. v. Repeat-O-Type Stencil Mfg. Corp., Inc., 123 F.3d 1445
(Fed. Cir. 1997).  Of course, here there is no sale, rather a license, and
there are indeed restrictions placed on the licensee, but I'm not sure those
differences would preclude the application of the implied license doctrine
to licensed software.  

Rather, I think the key issue is "uses of the [software] to which the
parties might reasonably contemplate the product will be put."  A clever
advocate may argue that the implied license granted by GPL is larger in
scope than the express license in MPL, in that, the latter is specifically
limited to the claims covered by the code as licensed by the patentee to the
licensee with those express restrictions you correctly point out in sec
2.2(d), whereas the implied license doctrine discussed above may grant a
patent license to the licensee to do much more than that, i.e. not just the
code as licensed by the patentee but also any "reasonably contemplated uses"
of that code, which may include creation and distribution of derivative
works since the terms under which the patented code is distributed permit
such activity.  Note that the implied license argument for MPL is much
weaker, and possibly not available at all, because there is express language
in the MPL stating exactly what is within the contemplation of the patentee.
Opposed to that are the licenses that are silent on the issue, GPL and
Apache as of now, where one may more easily argue that there has been a
broader grant of patent rights to the licensee.

Now, I'm not saying such a clever advocate would be meritorious, but I think
she would at least get past rule 11 (the ethical rule barring lawyers from
bringing frivolous arguments) and maybe even summary judgment (a judge
preventing an issue from trial because there is no way, even assuming all
the facts in favor of the advocate, to support the conclusion).  For
instance, one hurdle lurking for this line of argument is the reconstruction
vs. repair doctrine of patent law (reconstruction is not impliedly licensed,
but repair is).  Without going into that issue, let me just say that all of
this is not as predictable as we would all like it to be.  But, I will say
that there seems to be legitimate arguments that the free / open source
licenses that are silent vis-à-vis patents may actually grant their
licensees more rights to the licensor's patents than those licenses that
purport to address the issue.

With respect to amending your license, if you choose to do so, there are two
issues you may want to consider: (a) what claims are licensed, and (b) what
products are licensed.  Starting with (b), of course, amending the license
to require that patentee contributors grant a license to, not just the
version of the software as they distribute it, but all derivative works that
may be created of that version, would be a step towards a larger patent
grant.  For instance, the OSL does this.  

But, notice that the definition of Licensed Claims is limited in the OSL to
"patent claims owned or controlled by the Licensor that are embodied in the
Original Work as furnished by the Licensor."  This brings up issue (a), in
that if Developer, owner of patent claims covering A and B, modifies and
distributes the software to cover A, then Developer has granted, under the
OSL, a license to practice A with respect to the software as distributed by
Developer or any derivative work thereof.  What Developer has most likely
not done is grant a license to practice B in anything.  Therefore, a Third
Party modifying the Developer's version of the software to now practice A
and B, is likely not infringing the A claims, but is most probably
infringing the B claims.

So, to cover that situation, a license could hypothetically define Licensed
Claims as "patent claims owned or controlled by the Licensor that are
embodied in the Original Work as furnished by the Licensor or any Derivative
Work made in compliance with this license."  And, if one really wanted to
make the patent grant large, a license could hypothetically define Licensed
Claims as "any and all rights arising under U.S. or foreign patents or
patent applications, including rights arising through ownership, assignment,
license, equitable principles or otherwise owned or held by Licensor at any
time during the term of this License that are or may be infringed or
violated by any software licensed under this license [or any other license
approved as a Free or Open Source License by the Free Software Foundation or
Open Source Initiative]."  

You're issue with respect to not granting too broad of a license such that
proprietary developers could benefit is a good one, but precisely why FSF
and others argue that licenses such as MIT/ BSD/ Apache do not in fact do as
much to promote open source software as, say, the GPL does.  By not placing
derivative work licensing restrictions on licensees, the license, not a
patent clause within it, is actually the heart of the free rider issue that
concerns you.  But, that doesn't mean it can't be addressed.  For instance,
hypothetically, a license could address this by defining licensed claims as
"patent claims owned or controlled by the Licensor that are embodied in (a)
the Original Work as furnished by the Licensor or (b) any Derivative Work of
the Original Work (i) made in compliance with this license and (ii)
distributed only under this license [or any other license approved as a Free
or Open Source License by the Free Software Foundation or Open Source
Initiative]."

I know you mentioned the desire to lift language from someplace else, but
I'm not aware of any that meets your desires.  That's why I've suggested
hypothetical phrases here.

And, without pointing out the obvious, the larger the patent grant, the less
likely major patentees will adopt your license or participate in development
of software licensed thereunder.  However, I surmise such is not a large
issue for you and the Apache license.

Best,
--Dan

P.S.  If you're interested, I'll be giving a presentation titled "Patents
and Open Source: Potential Problems and Suggested Solutions" at LinuxWorld
in NY in January, where I'll address issues such as these.

Daniel Ravicher
Registered Patent Attorney & Associate
Patterson Belknap Webb & Tyler LLP
1133 Avenue of the Americas
New York, NY 10036
212.336.2826 direct
212.336.7900 fax
mailto:dravicher at pbwt.com
http://www.pbwt.com/

-----Original Message-----
From: Brian Behlendorf [mailto:brian at collab.net]
Sent: Monday, November 18, 2002 6:18 PM
To: license-discuss at opensource.org
Subject: MPL section 2.2, and patent grants on derivative works

Apologies if this is retreading old ground, and off-topic since it's not
about approval of a license, but we're working on the next version of the
Apache license and are looking at the patent language in various licenses.

I'm trying to tell if the patent grants by Contributors described in
section 2.2 apply to derivative works, either published by the same entity
as the Original Code, or a third party).  My read of 2.2 (IANAL, etc)
suggest that the patent grant *only* applies to the specific Contributor
Version, e.g. the Original Code plus the Contribution to which the patent
applies.  E.g.:

  Each Contributor hereby grants You a world-wide, royalty-free,
  non-exclusive license [...] under Patent Claims infringed by the making,
  using, or selling of Modifications made by that Contributor either alone
  and/or in combination with its Contributor Version (or portions of such
  combination), to make, use, sell, offer for sale, have made, and/or
  otherwise dispose of: 1) Modifications made by that Contributor (or
  portions thereof); and 2) the combination of  Modifications made by that
  Contributor with its Contributor Version (or portions of such
  combination).

What happens to a derivative work of the Contributor Version?  E.g., the
developers at Mozilla.org merrily commit the patch corresponding to that
Modification from said Contributor, and then make another commit, and
another, then make a release.  Given the language in the license, how does
that patent grant apply to this new work?  This new work is not the
"Modification alone", nor is it the "Contributor version", nor really "a
portion of such combination" - it contains the Modification, but even that
Modification may end up modified at some point.

2.2(d)(3)(i) appears to suggest that "third-party [who's a third party?]
modifications of Contributor Version" don't get that patent grant; neither
do those who "combine with other software".  Yikes!  That only further
confuses the situation.  Does putting Mozilla on a CD with some "other
software" and calling that a product constitute combination, and thus
lose the patent grants?

The root question for us (Apache) is, does a contributor grant need to
explicitly state that a grant of a patent license on a contribution
applies to derivative works of the resulting "contributor version".  If
that's the case, some say we need to limit the types of derivative works
it can apply to, since allowing any derivative work to carry the patent
grant would mean that anyone with a commercial product who wanted to use
that patent without paying could do so by pulling the right Apache code
into their own.  Limiting it to derivative works published by the ASF is
one option, published under any open source license is another.  But we'd
really prefer not to invent new language here, so I'm trying to understand
language like this in existing licenses.

	Brian

--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3

------------------------------------------------------------------------------
Privileged/Confidential Information may be contained in this message.  If you
are not the addressee indicated in this message (or responsible for delivery
of the message to such person), you may not copy or deliver this message to
anyone.  In such case, you should destroy this message and kindly notify the
sender by reply email.  Please advise immediately if you or your employer
do not consent to Internet email for messages of this kind.

==============================================================================

--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3