Subject: Re: IETF Patent Licenses are RAND
From: Eben Moglen <>
Date: Thu, 20 Mar 2003 22:34:39 -0500

On Thursday, 20 March 2003, John Cowan wrote:

  Eben Moglen scripsit:
  > The license is more
  > robust than that, and indeed--for reasons I've described elsewhere--
  > section 7 can actually provide important mutual defense against
  > assertions of supposed patent rights, but it can only do so if code is
  > licensed under it.
  Can you give us a pointer to these reasons?

A couple of people having asked off list as well, let me put the whole
matter in a nutshell:

Under US patent law, injunctions against distribution of infringing
goods are scarce as hens' teeth.  The adequate remedy is assumed to be
royalties by way of damages.  In the free software context, that means
independent developers and non-commercial distributors, both of whom
don't as a rule charge for their code, are pointless defendants to
sue.  The people who get hassled are the large commercial distributors
who make money and who therefore are potential sources of royalties.

Those enterprises, however, usually have substantial patent
inventories of their own.  Therefore, they don't pay royalties either;
they cross-license.  So let's suppose that (naming someone at random)
Microsoft has a patent that reads on some standard and wants to assert
its patent against, say, the Linux kernel's implementation of the
standard.  Suing individuals who don't charge won't work, so they go
to (again, picking a name at random) HP.  Under ordinary
circumstances, HP would cross-license on non-sublicensable RAND terms,
using its own patent portfolio to make a separate peace and leaving
the rest of us to hold the bag.  *BUT* not with section 7 of GPL.  If
HP accepts that RAND license, it has conditions imposed on it
incompatible with GPL, and therefore must stop distributing the Linux
kernel.  The only terms it can accept are GPL-compatible terms, and if
those terms are available to it, they are de facto available to all.
If it can't get a GPL-compatible license, HP has no choice but to
defend the infringement action, which it (unlike most of the rest of
us) has resources to do.

In this way, by preventing separate-peace deals among fat cats, GPL
section 7 actually ensures vigorous defense of patent assertions
against free software.  Unlike patent retaliation and "mutual defense"
provisions, which have some side effects of their own and which only
work against patent plaintiffs who also distribute free software, GPL
section 7 does more than discourage the assertion of claims; it
actually mobilizes the relevant resources by creating incentives for
well-funded potential *defendants* to do the right thing.

This approach to patent defense is--if you'll pardon a slight
professional joke--novel and non-obvious.  It's one of the subtleties
of GPL.  I'm not sure it's all we need; Richard Stallman and I are
still considering the full range of options for GPL v3.  But I think
it deserves careful consideration by those who are trying to decide
what license to use with one eye on the egregious problem of software

Best to all,

 Eben Moglen                       voice: 212-854-8382 
 Professor of Law                    fax: 212-854-7946       moglen@
 Columbia Law School, 435 West 116th Street, NYC 10027
 General Counsel, Free Software Foundation
license-discuss archive is at