[vox] More Ballmer FUD

[Karsten M. Self]

on Thu, Nov 18, 2004 at 11:47:54AM -0800, Robert G. Scofield (rscofield at afes.com) wrote:
> Bill Kendrick wrote:
> 
> >
> >*feh!*
> >
> > 
> >
> Yeah, but I'd like to go back to Rod's post.  Does anybody here think 
> that Linux is in any real danger?  Rod says it will be an interesting 
> year.  So supppose 228 patent infringement suits are filed.  Is there 
> any real threat?

First, you have to understand patents, patent law, infringement, and the
strategic use of patents.

Any given GNU/Linux system undoubtedly makes use of many patented
features.  Ditto *BSD, MacOS, VMS, MVS, WinXX, and any other operating
environment.  The patent space is so large, and computers (and operating
systems) so complex that it's difficult to write more than a few lines
of any code without treading on someone's patent.



Patent law is defined in Title 35, United States Code (aka 35 USC):

    http://www4.law.cornell.edu/uscode/35/

Nominally, patents are supposed to be limited to "novel and non-obvious"
inventions.  In reality, the US Patent and Trademark Office (USPTO) has
long been criticized for granting a large number of patents with
insufficient review, as well as vastly extending the scope of
patentability in the past thirty years, to include first software, and
in the past decade, business methods and other "inventions".  There's
even been discussion of patenting sports moves or techniques, though I
believe these are still only apocryphal.  See 35 USC 102.

http://assembler.law.cornell.edu/uscode/html/uscode35/usc_sec_35_00000102----000-.html

The problem with a grant of patent is that the real cost comes in
getting a patent declared invalid.  Regardless of merits, the process
typically costs on the order of one million dollars.  Generally, it's
far more cost-effective from a business standpoint to either pay up a
license fee (typically a few tens of thousands of dollars) or counter
with your own infringement claims (more on this below).  So even
low-quality patents tend to persist rather than be challenged.  While
analogies between real and intellectual property are dangerous, it's as
if a local government's land title office simply stamped "approved"
arbitrarially on 50% of all title applications, but required property
lines challenges go to court at an average cost of $1,000,000.



Someone infringes a patent when, as defined by statute, s/he:

    without authority makes, uses, offers to sell, or sells any patented
    invention, within the United States or imports into the United
    States any patented invention during the term of the patent.

Note that that's pretty broad.

Independent creation isn't a defense.  Unlike copyright, if you don't
copy someone else's work, but end up doing what they've independently
described, you can still be guilty of infringement.


Typically, patent actions are based on:

  - Licensing patents for a fee, to generate revenues.

  - *Not* licensing patents, and pursuing infringements, to gain an
    exclusive right to a particular business opportunity.

  - Cross-licensing or counter-suing for infringement, in response to
    another party's attempts to exercise their own patents against you.


All of which gets to how patents are typically used:  companies engaged
in a particular field file defensive patents.  These are patents which
either mark out a certain area in which they can operate exclusively,
*or* which provide ammunition to respond to another party's infringement
claims.

There are a few general scenario variants. 

 1. You're a company and you want exclusive rights in an area, *and* you
    own all or most key patents in the field.  Action:  don't license
    your technology.  Pursue infringement claims against those you
    suspect are infringing your patent (or are encroaching on your
    business).  This is relatively rare.

 2. You're a company with some key patents in an area, and you want to
    generate licensing revenue off of these.   Action:  license your
    technology to key industry participants.   The goal is actually to
    stay _out_ of court, so pricing is generally a small fraction of
    anticipated litigation costs, typically a few tens of thousands of
    dollars.  It's unlikely that any one licensee candidate will find it
    worth their time to challenge validity of the patent(s).  The
    extreme form of this operation is a "patent shop":  a company which
    has virtually no tangible assets other than a patent portfolio.
    Google for Samuel Lemelson for an interesting edge case.  Hint:
    he's got a wing of the Smithsonian named after him.

 3. You're a company interacting with others in an industry area.  Your
    competition holds one or more key patents in the area you're active
    in.  You wish to participate in the market without getting fleeced
    on licensing costs.  Action:  accumulate (file, buy, or otherwise
    aquire) key patent(s) in the area.  When presented with a licensing
    claim, counter-claim with infringement of your own portfolio.
    Objective is to cross-license patents such that you and your
    competition effectively pay no licensing fees.  Note:  see Red
    Queen, MAD.

 4. You're Microsoft, and you're worried about GNU/Linux.  At present,
    the strategy isn't clear, and direct enforcement actions have been
    quite limited.  However, stick-waving and FUD might be effective in
    swaying some markets and corporate decisionmakers for the time
    being.  There are a few other aspects to worry about, however.  The
    situation does more-or-less appear to be all-or-nothing, however, so
    ultimately you've got everything to lose.



The problem with Free Software is generally seen as:  Free Software
developers have few or no patent holdings, and lack the resources to
fight an infringement claim.  The patent system effectively acts as a
damp towel (or dam burst) on any potential Free Software development.

The reality is a bit more complicated.

In the patent space, there are basically two players:  IBM, and everyone
else.  Actually, there are a few big players, of whom IBM is the
largest, with a patent portfolio of tens of thousands of patents
generating well over $1 billion in annual licensing fees (this after a
big push by IBM in the late 1990s to increase revenues)[1].  Other major
holders include many of the major US, Japanese, and European technology
companies.  Interestingly, until the late 1990s, Microsoft itself was a
bit player, and still has only a modest portfolio.

If you actually run the numbers (and they're available from the USPTO,
I've dug them out in the past), it turns out that more than half of all
patents are held by parties with only a small number of patents
(certainly less than ten, possibly fewer) apiece.  Which raises some
interesting possibilities if these individuals might be convinced to use
their patents in some collective way.  Which also raises some
interesting possibilities to the US DoJ's antitrust department (patent
pools are seen as highly suspicious by way of possible antitrust abuse).


So in the Free Software space, it turns out that:

  - Individuals and small companies actually do have relatively
    significant patent holdings, in aggregate.  Some of these patents
    are themselves highly valuable.

  - There are major players with a strong interest in Free Software
    (IBM, Intel, HP, Toshiba, Sun, Novell, etc.), who themselves have
    major patent portfolios.

  - Microsoft has only a modest patent holdings.  It's got problems with
    perceived antitrust behavior in the US, Europe, and Asia.  It's got
    a large number of corporate enemies in the US and elsewhere.  Most
    likely, we're going to see continued proxy wars (Wang vs. Netscape,
    Caldera/SCO vs. IBM, etc.) rather than direct engagement, though
    Microsoft seem to be getting bolder (their relationships with
    Caldera/SCO are both plainly evident and openly acknowledged)[2].


When I looked at this issue some years ago, around 1998, I basically saw
a couple of possible scenarios:

  - A Free Software patent pool.  This raises a number of issues,
    largely organization and antitrust.  There's the participation
    incentive question (why should I participate, how should I structure
    my own risks), particularly in the bootstrap mode.  One modern
    morphing of the concept appears in several minor Free Software
    licenses' patent licensing clauses, along the lines of mutual
    self-defense and/or poison-pill pacts.

  - "Patent Patrons":  the idea that eventually there would be
    sufficient interest in Free Software among major corporate players
    that the patent problem would, if not disappear, at least largely
    fade into the background noise level.  
    
    
This latter appears to be what's largely taken place, and far beyond any
dreams I might have had six years ago.  The main problem with this
scenario is that coverage is ad hoc and limited, and subject to
"flipping" of key players:  their deciding that, no, actually, GNU/Linux
*isn't* in our corporate interest today.   Ad hockery cuts both ways
though, and makes the risks of embarking on patent litigation high for
the potential plaintiff, due to uncertainties:  will the Big Guns get in
on this one or not?

For any given GNU/Linux user, the risks of enforcement actions are
relatively small.  The risk is *largely* one of Microsoft (or a proxy)
acting in what can really only be described as a shake-down operation:
pay us protection money, or use our products, or else.  The likelihood
of this happening is mitigated by a number of other factors, starting
with IBM and other major tech players with a strong interest in
GNU/Linux, who might retaliate in turn against Microsoft.  Presently,
the result appears to be a stand-off.



Upshot Q & A:

    Q: Does GNU/Linux embody third-party patents
    A: Certainly.

    Q: Does it do so to any greater extent than other software products.
    A: No.

    Q: Does the open-source nature of GNU/Linux pose a greater risk to
       patent enforcement actions than closed-source software?  
    A: Possibly.  But more likely, the most severe patent risks come
       from Microsoft's own self-perceived corporate vulnerability to
       GNU/Linux expansion.

    Q: Are there parties _other_ than Microsoft who might seek to
       collect on patentent licenses in Linux or other Free Software
       projects?  
    A: Almost certainly.  However, save Microsoft (and proxies as
       previously discussed), most of these would operate as rational
       businesses:  their license claims will be relatively small, the
       requests will be against companies capable of paying, and the
       concern will be at keeping transaction costs in line with
       revenues.

    Q: Am I as a personal user at risk?  
    A: Possibly, but low.  Remeber:  in the US, pretty much anyone can
       sue anyone for anything.  Even frivolous lawsuits can be
       expensive in terms of time, costs, and sanity.

    Q: Is my company using/making/selling GNU/Linux-based solutions or
       services at risk?  
    A: Possibly, but likely low.  I'd keep an eye on Redmond.  And see
       above regards suits.

    Q: What can I do to reduce risks?  
    A: In the fight-fire-with-fire dept: aquire patents, or become good
       friends with those who have them.

    Q: Can the patent system be reformed?  
    A: Yes.  And with sufficient initial thrust, pigs will fly.

    Q: You don't sound very optimistic.
    A: That wasn't a question, but the answer is:  there's a very
       strong, entrenched, and self-reinforcing network of interests
       keeping the patent system as it is.  Few people outside a very
       small community have any idea of the problems, let alone the
       solution.  And there's a very strong likelihood that "reform"
       would be coopted by those who benefit from the current system.

    Q: What's your own preferred fix for the system?
    A: Make the USPTO refund registration fees to patentholders whose
       patents are invalidated, *and* pay a substantial fraction of
       costs to the challenger.  This would turn the current
       rubber-stamping system into a large governmental revenue
       liability.  You've got to be careful about making it too easy for
       large companies to stomp on individual inventors, who, curiously,
       are one *very* vocal support group for the current system.

    Q: That doesn't address the patentability of software, business
       methods, sports techniques, etc....
    A: No.  But it does force the USPTO to be accountable for the
       letters patent it does grant, a very good first step.  Further
       reforms could follow.  Alternatively, if Free Software supporters
       could figure out how to do a GPL-like IP jujitsu move using
       patent law, we might see either a neutralization of the threat,
       or a spiking of the system sufficiently that a critical mass
       necessary to change the current patent system would emerge.

    Q: Really, what's the answer?
    A: 42.


Peace.

--------------------
Notes:

1.  The $1 billion plus does *not* include, to the best of my knowledge,
    software-based licensing.  This is based on conversations with
    several IBM personnel, and the rationale was essentially that
    opening *that* Pandora's box was deemed to be too great a corporate
    risk, despite Legal's salivation over what is undoubtedly a healthy
    revenue potential.

2.  Microsoft have engaged several times in what can only be called "war
    by proxy" battles.  Wang vs. Netscape was a patent-based claim
    against Intel for a trivial but fundamental bit of browser
    capability.  The patent was invalidated on review, though at some
    nontrivial cost to Netscape.  The Caldera/SCO vs. IBM case is a
    second.  The usual proxies are generally washed-up technology
    companies with little other than an IP claim (patent, copyright),
    and an ability to be swerved by Microsoft.  I don't think we've seen
    the end of this, and would look with some concern at companies with
    much to lose, among them, Sun and Unisys.

-- 
Karsten M. Self <kmself at ix.netcom.com>        http://kmself.home.netcom.com/
 What Part of "Gestalt" don't you understand?
  Information is not power after all: Old-fashioned power is power. If you
  aren't big industry or government, you have very little power. Once they've
  hacked the electronic voting system, you'll have no power at all.
  - Robert X. Cringely
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://ns1.livepenguin.com/pipermail/vox/attachments/20041122/ec0a7445/attachment.bin