Defense against Patents

[Eben Moglen]

Larry,

Yes but no.  

Let's begin, before disagreeing, by agreeing.  I haven't said that the
measure represented by GPL 7 is enough; that's not a conclusion I am
yet prepared to draw.  Deciding how and to what extent to invoke other
forms of defense against patent assertions is very important in the
further development of free software licensing.  I don't mean to rule
out other measures; what we intend to do in GPLv3 is still uncertain.
But you underestimate the problems of the patent plaintiff, and also
underestimate the contribution GPL 7 makes to making those problems
worse.

We need to differentiate between two kinds of patent plaintiffs:
roughly, Microsoft and everyone else.  Everyone else wants royalties;
Microsoft wants to stop free software.  In fact, it is Microsoft that
has people worried.  But Microsoft's choices are no better than the
other fellows'.  We need to take both branches of the analysis.

Those who want royalties don't sue independent developers because the
costs of suit exceed by orders of magnitude the available gains.
Those who want to stop free software don't sue independent developers
because bringing thousands of Goliath v. David actions is bad public
relations for a monopoly.  You and I have both heard the relevant
Microsoft strategist, in a confidential industry forum, disclaim any
intent to sue "garage inventors."  It's not trust, but an accurate
estimation of the inevitable consequences, that causes me to believe
her statement.  So far as equitable relief goes, an injunction against
the distribution of free software is useless: ask the movie industry.
The studios spent more than $12 million trying to enjoin the
distribution of DeCSS; they won almost every case, and the software is
ubiquitous.  Those who want licensing income don't need to spend money
getting no royalties, and those who want to stop free software know
they can't do what Jack Valenti and Michael Eisner can't do either.
No, I don't think injunctions are going to be awarded in US courts
against free software products in patent cases.  But it's an empirical
question: how many preliminary injunctions in patent cases did you see
awarded in the district courts last year?  That just isn't how it
goes.

If you want licensing revenue, suing banks, insurance companies and
manufacturing firms may be a good idea, but if you're Microsoft, it's
a terrible decision.  Those people are your customers, and if you sue
them you make them angry at you.  Ain't gonna happen.  And collecting
royalties from those who use products infringing patents, as opposed
to those who sell them, is theoretically possible, but only just.
When was the last time you saw a patent-holder trying to collect
royalties from consumers?  

May I whisper something in your ear?  Most software patents are
invalid.  You know it, I know it, the patent-holders know it.  Patent
assertion in the area of software is a bluffers' game, and it works
because parties cross-license rather than challenging.  What GPL 7
does is to make the likelihood of challenge much higher, because the
opportunities for settlement are artificially constrained.  That means
the patent-holder faces a different payoff matrix, and plays the game
differently at the margins.  There are nuances and refinements to this
analysis, not all of which I want to talk about publicly, but I stand
on the proposition--which is not all that strong or surprising,
really--that GPL 7 has some subtle effects on the environment for
potential patent claimants, and that those subtle effects have been
important in preventing assertions and litigations.

Which brings us to the real point: I'm not just guessing, I've been
there.  Even specious software patents weren't invented yesterday, and
we've got a little experience with what happens when the game plays
out.  I can think offhand of eight GPL'd programs, three of them
stalwarts of everyone's operating environment from day to day, that
have been the subject of someone's intended patent holdup at one time
or another.  Each patent-holder presenting a claim has found that the
usual "roll over and license" approach wouldn't be taken by the big
boys, because of GPL 7, and each holder eventually thought better of
trying to sue.  I didn't learn that GPL 7 protects against patent
actions by reading the license; I'm not that clever.  I learnt it by
watching what actually happened in the world.

Now, once again, I'm not contending that more strenuous forms of
defense are a bad idea.  They may work; they may not.  I quarrel with
no one who wants to invent a better mousetrap.  I simply say that our
mouse-deterrer has kept the rats away so far, and it is worth a second
look by those who want something practical right now.  

Best to all,
Eben


On Friday, 21 March 2003, Lawrence E. Rosen wrote:

  Eben,
  
  Your analysis of how the GPL might protect software from a patent threat
  is not realistic.
  
  You hypothesized, solely for discussion purposes, that Microsoft has a
  patent that reads on some portion of Linux.  You suggested that
  Microsoft would not go after the open source programmer (his pockets are
  too shallow) but would go after, say, HP instead.  Because of the GPL §
  7, you concluded, HP would not be able to construct a private settlement
  but would have to defend everyone's right to practice the patent.
  
  But why would Microsoft go against HP and risk HP's funding litigation
  to invalidate the patent?  That is not typically how such patent
  disputes resolve themselves.  Patent holders often try to assert their
  claims against little guys first.  That way they can get quick
  settlements that establish a de facto legitimacy to the patent without
  risking a big-bucks defense.  They litigate slowly up the food chain,
  not down it.  
  
  Alternatively, Microsoft could forget a lawsuit against HP and proceed
  against HP's customers directly -- the banks, insurance and
  manufacturing companies that infringe through their *use* of patented
  technology.  Such companies don't usually have technology patents that
  they can use as bargaining chips.  Also, as you well know, there is no
  indemnification or warranty of non-infringement in open source licenses.
  Merely chilling the acceptability of our software in the eyes of our
  customers would be a dreadful defeat for us.
  
  And why would injunctive relief not be appropriate, even against little
  guys?  When arguably infringing software is distributed for free there
  is no reasonable measure of monetary damages.  As you know, that is one
  of the criteria courts use to judge the appropriateness of injunctive
  relief.  If you can knock out the key developers of an open source
  project because of the threat that they will lose their houses and their
  software (unrealistic though that threat might be to those of us who
  understand real litigation in real circumstances), wouldn't that suffice
  to chill the use of patented technology?
  
  I'm sorry to say that the GPL's defense against the patent threat is
  weaker than we can afford.  We need something stronger.
  
  /Larry Rosen
--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3