On the licensing of software patents [LWN.net]

Unless it is changed before adoption, the proposed W3C royalty-free patent policy will allow "field of use" provisions. Patented technologies which are included in a W3C standard must be licensed for royalty-free use - but only for implementations of the the standard itself. Owners of patents can still require license payments for any other use of the technology.

What this means, of course, is that, if a W3C standard contains patented technology with "field of use" restrictions, no implementation of that standard may be distributed under the GPL. The GPL does not allow that sort of restrictions. Free implementations of such standards can be distributed under BSD-style licenses, so it remains possible to implement the standard in free software. But the range of that freedom has been restricted somewhat.

These terms make an interesting contrast with another form of royalty-free patent licensing. Companies like Red Hat and FSMLabs have licensed their patents for use in free software - but only for software licensed under the GPL. BSD-licensed implementations are not covered by these patent licenses.

If these trends continue, the proliferation of software patents is going to bring about a partial partitioning of the free software ecosystem. The two types of patent licensing are, essentially, allergic to each other, and can not be mixed. This is not a new situation - mixing free software with different licenses can be problematic even without the additional complication of patent issues. But adding in incompatible patent licensing creates new and dangerous problems.

Software patents may well turn out to be one of the more potent weapons against free software in general. Patent infringement lawsuits can be filed against any user of the allegedly infringing software, not just its developers or distributors. A couple of high-profile examples of companies being dragged into court for using a free program would serve to create a great deal of fear, uncertainty, and doubt among all free software users - even if the patent suits are eventually tossed out. The free software will have to step carefully when implementing algorithms covered by patents - and that may well not be enough.

to post comments

This is not just a GPL issue

Posted Jan 2, 2003 20:53 UTC (Thu) by stevenj (guest, #421) [Link] (6 responses)

"Free implementations of such standards can be distributed under BSD-style licenses, so it remains possible to implement the standard in free software."

You're missing the point, I'm afraid. In fact, such use-restricted standards cannot be implemented as free or open-source software under any license, BSD or GPL or otherwise. A field-of-use restriction would conflict with both the FSF's Free Software definition and the OSI's Open Source definition.

The GPL issue is a red herring. Yes, the GPL is also legally incompatible with such a patent, whereas (e.g.) BSD is not. BSD code can also be incorporated into proprietary software, but this doesn't make MS part of the free software world because it uses some BSD code! What matters here is what we want, not what we are legally compelled to do.

If you care about free (as in speech) software, you're out of luck with such patented standards. The "free enough" crowd strikes again.

(As a concrete example, field-of-use restrictions could lead to a legal minefield, depending on how they are worded, because W3C standards have a broad influence beyond their ostensible field of web clients and servers.)

This is not just a GPL issue

Posted Jan 3, 2003 5:37 UTC (Fri) by ksmathers (guest, #2353) [Link] (5 responses)

It may be a problem for software that is trying to comply with the open source definition, but (my IP lawyer tells me) that it is not a problem for GPL. GPL is silent on the issue of patents. It does state that any license under which the software is distributed must be at least as free as the GPL, but that doesn't inherently restrict GPL'd software from implementing patented algorithms. Unlike copyright, software patents are not recognized worldwide, must be filed separately in each country where the ideas are to be protected. GPL in my opinion correctly ignores these different interpretations and says only that the source code must be freely distributable under no more restrictive conditions..., etc.

Since one is a restriction on copying, and the other is a restriction on use, they arguably are non-overlapping concerns. A free implementation of a protected idea should be perfectly compatible with GPL.

This is not just a GPL issue

Posted Jan 6, 2003 0:00 UTC (Mon) by mbp (subscriber, #2737) [Link] (4 responses)

The point is that it may make it impossible, or difficult, to distribute and use GPL'd software in practice.

Software patents are a terrible problem for the industry as a whole, and for free software in particular. (See for example Phil Karn's comments: <http://people.qualcomm.com/karn/patents/patents.html>) A supposed open standards body has no business encouraging their further use, or their incorporation into standards.

This is not just a GPL issue

Posted Jan 16, 2003 14:24 UTC (Thu) by jdthood (guest, #4157) [Link] (3 responses)

Not so.

Let's invent an example to make this clearer. Suppose a web standard is introduced that involves a patented algorithm for encrypting information in transit. Company C owns the patent, but agrees to license the patent royalty-free for use on the web.

Hacker H1 writes a program that implements the algorithm in a browser plugin and publishes it under the GPL.

User U1 uses this plugin to access a secure website.

So far, no one has broken the law. U1 used the algorithm within the terms of the patent.

Now, Hacker H2 modifies H1's program to encrypt traffic between a database client and server. He releases his program under the GPL.

U2 uses H2's program to access databases securely.

I suspect that either H2 or U2 has broken the law. However, that does not change the fact that the original article got it wrong. LWN wrote:
    ... if a W3C standard contains patented
    technology with "field of use" restrictions,
    no implementation of that standard may be
    distributed under the GPL. The GPL does not
    allow that sort of restrictions.
That is wrong. Such software *can* be distributed. There is no incompatibility between the GPL and field- of-use restrictions. The GPL disallows the introduction of restrictions *into the copyright license*, but says nothing about restrictions which may exist independently of that license.

This is important. If I am write then I hope that LWN.NET will publish a correction at some point. Otherwise, someone please correct me.
--
Thomas Hood

This is not just a GPL issue

Posted Jan 16, 2003 14:39 UTC (Thu) by corbet (editor, #1) [Link] (2 responses)

"The GPL disallows the introduction of restrictions *into the copyright license*, but says nothing about restrictions which may exist independently of that license."

What I wrote is based on section 7 of the GPL:

If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program.

The GPL disallows restrictions on derived works. Field of use restrictions are a restriction on derived works. I believe that what I wrote is correct.

This is not just a GPL issue

Posted Jan 16, 2003 22:24 UTC (Thu) by jdthood (guest, #4157) [Link] (1 responses)

Hmm.

> For example, if a patent license would not permit
> royalty-free redistribution of the Program by all
> those who receive copies directly or indirectly
> through you, then the only way you could satisfy
> both it and this License would be to refrain
> entirely from distribution of the Program.

You are right that the GPL does say something about
restrictions that exist independently of the license.

But I maintain my disagreement ... for now. The way I read it,
section 7 is redundant or nearly so: it just emphasizes
that if you can't legally (re)distribute the work under
the GPL then you can't (re)distribute it at all. But
this already follows from the rest of the license. I
suppose the section was added in order to forestall
(by making its illegality more explicit) an obvious dodging
tactic.

The crucial question is: DO the conditions imposed by
field-of-use patents somehow "contradict the conditions of
this License"? If so, then the work cannot be distributed
under the GPL. If not then it can. Let us try to answer
this question.

You say that "the GPL disallows restrictions on derived works"
and reason that because a field-of-use patent imposes
restrictions, implementations of such patented ideas cannot
be distributed under the GPL.

However, I think that there is a slip in the reasoning here.
The GPL disallows the introduction of more restrictive terms
into the copyright license of the derived work to be
distributed. A field-of-use patent imposes restrictions quite
independently of the copyright license. Its authority
derives from an entirely different law. Obviously the GPL
cannot disallow *these* sorts of restrictions. If it did,
then nothing at all could ever be published under the
GPL, because all software distribution is restricted by some
part of the law. The law says (I presume) that I cannot
distribute software that contains child pornography. This
is a restriction, but surely it is not a restriction that
conflicts with the GPL.

Regards,
--
Thomas Hood

This is not just a GPL issue

Posted Jan 16, 2003 22:45 UTC (Thu) by jdthood (guest, #4157) [Link]

Just thought it might be useful to add a link to a posting
by Bruce Perens on this subject in debian-devel mailing list.

http://marc.theaimsgroup.com/?l=debian-user&m=104138016327196&w=2

He doesn't see a problem with GPL implementations of
field-of-use patent encumbered algorithms, so long as
the relevant code is not used to do anything outside
the permitted field.

On the licensing of software patents

Posted Jan 5, 2003 23:56 UTC (Sun) by mbp (subscriber, #2737) [Link]

One day, the disciple said unto the master: "Master, there are so many
free software licenses with so many different requirements that nobody
can possibly remember them all. I shall design my own set of unified
free software licenses to replace all the others, and thereby set us
all free!"

The master immediately slapped the disciple.

"Master, why did you strike me?"

The master said nothing, and the disciple went away.

The next day, the disciple returned to the master, but before he could
say anything, the master immediately slapped the disciple.

"Master, why did you strike me again?"

"I did not strike you again. Yesterday, I struck you with my left
hand. Today I strike you with my right hand. Tomorrow I shall kick
you."

The disciple was suddenly enlightened.
-- Tim Lesher

(Not really quite the same issue, I know. I just like it.)