Subject: Re: patent trolls and X-licensors
From: Thomas Lord <lord@emf.net>
Date: Tue, 06 Jun 2006 12:09:27 -0700

Ben Tilly wrote:
> On 6/5/06, Stephen J. Turnbull <turnbull@sk.tsukuba.ac.jp> wrote:
>
>> For abolitionists, the exercise that Tom is conducting should be
>> viewed as an *essential* first step in demonstrating that patents
>> *can't* work, for example, and pointing to problems in the *existing*
>> system is not an answer to the questions he's raising about a possible
>> alternative.
>
> Tom has said a tremendous amount, and if I wade through it I'm
> unlikely to draw the right conclusion about what part of it you're
> thinking of.  Which exercise of Tom's are you talking about?

It's now spread over several messages as other people
added good criticisms and suggestions.  So here is
a "quick" summary.  It's a work in progress.


* The Software Patent Disease

  I think we all agree, more or less, about the symptoms of the
  disease:

  The rules for patenting software began to change in the 1970s
  and underwent substantial liberalization in the 1980s and
  1990s.  Anecdotal evidence (e.g. {rms}) and some statistical
  evidence (e.g., {Bessen and Hunt}) suggest that the incentives
  created in the 1980s and onward led to an increase in
  strategic software patent portfolios (e.g., defensive patent
  stables, patent trolling) which was counter-productive at
  generating R&D in some sectors.  Anecdotal evidence (e.g.,
  {rms}) suggests that the broad scope of patents and their
  sheer numbers (a result of strategic patenting) makes
  programming a minefield in which it is essentially impossible
  to write a substantial original program without risking
  infringement on some patents that it would be all but
  impossible to discover by search.  Finally, I note a
  particular variation on strategic patents: the use of patents
  to enforce an arbitrary and trivial technical choice as a de
  facto industry standard, creating a form of platform exclusion
  unwarrented by the incentive goals of the patent system.

* What kind of cure is needed?

  Anti-software-patent absolutists (of which anti-patent
  absolutists are a subset) argue for the complete elimination
  of software patents.  Some of the arguments are from the
  perspective of practicality -- the current system does more
  harm than good and there's no way to fix it.  Some of the
  arguments are more metaphysical -- e.g. programs are math,
  pure ideas, and there should be a clear subject exclusion of
  them from patents.

  I note that (a) it is easy to make a strong economic case for
  patents in general, (b) it is very difficult to define a
  "software patent" as distinct from other kinds in a legally
  robust way -- part of the reson we went down this slippery
  slope in the first place, and (c) current abuses and failures
  in the area of software patents can arguably be attributed not
  to the fact that software can be patented, but to the
  particular rules of how the patent system operates.

  So, in contrast to anti-patent absolutists, I propose simply
  changing the rules.  And in making changes, I propose changes
  that would effect *all* patents equally, with no need to
  legally define a special class of "software patent".

  In the discussion on the FSB mailing list I proposed a
  specific set of rules.   Stephen proposed a different
  rule change.   Other criticisms were made and I expanded
  and clarified my rules.   So here is a summary of the
  current state of proposed rule changes -- a synthesis of
  much of the preceeding discussion.   There are some original
  tweaks to previously discussed rules here, too.

  Rules changes may be the subject of advances in the
  constructions of the judiciary (where they are simply clearer
  readings of existing laws) or the subject of new legislation.
  I've mostly avoided (and continue to do so) the question of
  which side of that line each of my proposed rule changes fall
  on.


* The Negotiating Power of Untested Patents and Liability for
  False Demands

  Consider an agressive holder of a large strategic patent
  portfolio who notifies a third party that they are infringing
  on each of 10 patents.

  Let's stipulate that in court the defendant would challenge
  either the validity or the applicability of each patent and
  receive judgements in his favor.

  Nevertheless, going to court and making those challenges is
  likely to be difficult, time consuming, and expensive.  It is
  also risky because, prior to receiving that judgment, a court
  may be persuaded to issue a temporary injunction against the
  defendant.

  In this calculus, if the settlement demand of plaintiff is low
  enough (and it can still be quite high) it may be rational
  business for the defendant to settle.

  This creates incentive for the practice known as "patent
  trolling": the acquisition of collection of strategic patents
  for the sole purpose collecting as many settlements as
  possible.  Patent trolls are characterized by the absense of
  credible effort to practice a patent -- they work instead to
  collect revenue by issuing demands and filing court cases.

  I propose several mechanisms for removing that incentive from
  the patent system the first of which is plaintiff liability
  for issuing unfair demands.  This is tricky to formulate in a
  fair way but I think it can be done.

  Plaintiff's should be liable for some or all of defendant's
  costs, plus punative damages, in three cases:


        1) The plaintiff issues a demand letter to the defendant
           without first securing and presenting, from a
           licensed panel of experts authorized by act of
           Congress, evidence of (a) the validity of the patent
           and (b) the applicability of the patent in this case.
           The findings of these experts are not legally binding
           but they must be informative to the defendant and
           consistent with a case plaintiff would bring to
           court.

        2) The plaintiff withdraws the demand upon receiving a
           response from defendant who has himself secured
           and presented refuting evidence from a licensed
           panel of experts where that refuting evidence
           makes use only of information that was available to
           plaintiff's experts.

        3) A court dismisses plaintiff's complaint
           basis of evidence and reasoning clearly articulated
           in defendant's response to plaintiff's demand
           with an additional finding that that nothing
           in plaintiff's presentation could have been
           reasonably expected to overcome defendant's
           response to the demand.
          



  The extent of the liability is determined as follows:

  a) all reasonable legal fees directly relating to court
     proceedings initiated by the plaintiff

  b) plus a penalty of $5,000 in case (1), $100 in case (2),
     and $500 in case (3)

  c) plus, in case (2), an amount, not to be less than $750, but
     which is otherwise the lesser of an "apology amount"
     specified by plaintiff in the initial demand or defendant's
     bona fide legal and licensed expert expenses in responding
     to plaintif's demand

  d) plus, in case (3), the full reasonable cost of plaintiff's
     legal and retained expert expenses in response to the
     demand letter and in court


  The defendant has some obligations for the liability to apply.
  Prior to sending a demand letter, plaintiff's licensed experts
  may request information from defendant about details of
  defendant's operations that pertain to a potential
  infringement claim.  Defendant may demand the licensed experts
  to enter into an NDA which excludes disclosure to the public
  and to plaintiff but defendant must (for the liability to
  apply) disclose any information it will later use as defense
  in court.  If the plaintiff's experts report (concealing NDA
  information) to plaintiff "you don't have a case here" -- that
  must be disclosed to the potential defendant immediately.


  Those liability rules alone aren't enough to fix the
  "negotiating power of bogus" patents -- it is a half measure.
  The problem is that in the current system some "bogus" patents
  may still be found valid and applicable.  Additional rule
  changes are needed which complement the liability and bring
  our notions of validity and applicability more in line with a
  reasonable social policy notion of what makes a patent "bogus"
  in some circumstances.   The next several rule changes address
  that.
          
 

* Scope Limited by Dilligence

  Patents grant an exclusive right to exclude others from
  certain uses of an invention. {Scalia}

  The patent system is Constitutionally authorized for the
  express purpose of "promoting progress in the useful arts and
  sciences". {US Constitution}

  It seems to me that patents are meant to exclude third parties
  from interfering with specific diligent activities by the
  patent holder which may result in progress in the useful arts
  and sciences.  If a third party's activity does *not*
  interfere with the diligent persuit of progress by the holder,
  then no infringement has occured.

  For example, prior to the invention of the World Wide Web a
  patent may have be obtained by an inventor diligently
  developing BBS ("bulletin board system") software for use with
  ad hoc dial-up networks.  The inventor might have obtained a
  patent for a system of collecting user registration data over
  a network and using that data to generate statistics on
  subscriber usage patterns.  We can stipulate (though not all
  would agree) that this was a truly original invention at the
  time.  The patent holder practiced the patent for several
  years, building a business selling BBS software.  As the
  Internet took off, the market for BBS software diminished and
  the patent holder eventually shut down operations.  The patent
  holder made no further practice of the patent.  Meanwhile, the
  World Wide Web was created.  User registration and statistics
  gathering became a common-place feature of much web software.
  At no point to the patent holder exercise his patent rights to
  further that development.

  Now it may be the case that a strict reading of an example of
  today's Web software will show that it satisfies every (very
  abstract) claim of the original BBS patent.  Nevertheless,
  none of the third parties writing web software that satisfies
  these claim has in any way interfered with diligent efforts by
  the patent holder to make progress in the useful arts and
  sciences on the basis of his invention.  No infringement has
  taken place.

  In short, it needs to be clearer that the scope of a patent is
  limited by how the patent holder attempts to use it.

  That scope can still be quite broad.  If very early on in the
  web the original patent owner had recognized his invention in
  some early systems (and noticed those site owners), or had
  recognized the relevance of the patent to the web on his own
  -- he might have tried to promote his patent.  He might have
  written to various companies suggesting they implement such a
  feature, offering them reasonable terms.  He might have tried
  to start a company that would build a web server module
  implementing the feature.  He might have diligently practiced
  the patent in any number of ways.  But if he just left it
  sitting on a shelf and then, at a late date, said "by the way,
  all you guys now owe me a lot of money" -- that is outside of
  his "exclusive right to exclude".

  The scope/diligence rule is particularly important to software
  patents because of the unusual role that abstraction plays in
  computer programming and in the language used to describe a
  computer program.  Very different inventions, being deployed
  for very different business purposes, can sometimes
  nevertheless both be described as "the same" in the abstract
  language of a patent.  It makes no sense to enforce a patent
  on one of the inventions against the other unless the other is
  *in fact* interfering with the patent holder's own practices
  of his patent.  The patent system isn't meant to be a lottery
  system that pays off when it is accidentally discovered that
  two unrelated inventions happen to fall under a common
  abstract description.


* Term Limited by Practice and Uniqueness -- De Facto Standards

  Sometimes, as with video codecs, patent law is currently being
  applied to protect a piece of software in order to create a
  monopoly on a de facto public standard rather than to reward a
  genuinely precious insight.

  Each new video codec does represent some non-trivial
  investment in R&D but, within a given generation of comparable
  codecs, no one of them represents a substantial advance in
  the art of designing codecs.

  Concurrently, because of how video codecs are used on the web,
  the main market advantage of one code over another is "which
  one do programs from Microsoft support by default?"

  A publisher of video, on the net, if they want to reach a
  large audience, pretty much must choose a video codec that
  is bundled with or similarly endorsed by Microsoft.

  Now let's step pack and examine patents.  None of the current
  video codecs is especially "first to market".  With the
  possible exception of some bleeding edge thing I haven't heard
  about, none of them are unique or remarkable.  If someone
  tells you you may not use codec X for some isolated task then
  pick coded Y and you're just as happy.  Worst case, put some
  grad students on the problem for a few months and get codec Z.

  How many ways are there to fold a piece of paper small enough
  to fit in your shirt pocket?   Is it really socially useful to
  allocate exclusive rights of exclusion to any one of them?

  Patents are being used in this and areas like it not to
  protect an invention -- but to protect a cartel which aims to
  create a de facto public standard.   I don't think patent
  protection should extend that far.   Informally, the act of
  trying to create a public standard should imply a public
  license to the patent.   How can we formulate that in law?
  I propose a four point test:


    1) Are competing designs to a given invention relatively
           inexpensive to create relative to the market size for
           products that encorporate the invention but add
           significant additional functionality?  Do competing
           inventions exist in abundance?  Were some created
           without the incentive of patents or is it easy to
           imagine circumstances in which some would be?

    2) In a product incorporating the invention, is the
       choice among competing inventions arbitrary from a
       technical perspective?

        3) Do the recognition of the need for the invention, and
           a general appreciation of the design space precede
           the invention?  Can dedicated professionals in the
           field, simply through moderate systematic work,
           satisfy the well understood need?
 
        4) Is the patent holder's business model largely to
           secure licensing from a dominant share of those in
           need of some invention offering similar capabilities?
           Is the business model based not on the advantages of
           the invention over its competitor but on its market
           share?  Is the reason a licensee might choose
           invention A over invention B primarily that other
           licensees made the same otherwise arbitrary choice?

  If the answer to all four questions is yes, the patent is
  advanced immediately to the public domain. 


* Validity Limited by More Informed Judgements of Obviousness

  A preponderance of testimony of domain experts regarding
  the obviousness of a need for a patented invention and the
  details for the invention should be determinative on the
  question of obviousness concerning the validity of a
  patent which has been issued.


* Term Limitations

  I don't have a detailed proposal here.   Many of us agree
  that 20 years is obsurd for some forms of invention.  For
  example, 20 years can greatly exceed the lifetime usefulness
  of some software inventions.

  I note that in the waning years of a software invention
  of lasting usefulness -- the RSA patent -- RSA's pricing
  schemes reflected that even short-term patent protection had
  signifcant economic value for a software product.  (The price,
  an anecdote on the FSB list reports, went *up* as the date of
  public domain release approached!  Licensees evidently wanted
  to buy the license in anticipation of the impending
  public-domain date -- so they would be well prepared on that
  date.)

  Perhaps it would be useful to permit patent holders to make
  binding declarations to the Patent Office that certain patent
  will be released to the public domain in some time short of
  the 20 year term.


* Public Review

  Stephen Turnbull has raised the idea, on the FSB mailing list,
  of a fundamental shift in the patent issue process.  If I
  understand correctly:

  A patent application is privately reviewed by the PTO and may
  be rejected.  If accepted, the acceptance is provisional.
  Provisional acceptance means that the application is publicly
  disclosed but does *not* mean a patent is issued.

  A period of public commentary follows provisional acceptance.
  Following this period, the PTO may or may not grant the
  patent.  In the case of acceptance, the public commentary
  remains attached to the patent in the PTO's publicly
  searchable database.


-t

Footnotes:

  I owe these but the "{terms}" should make them easy to find.

  Various writers on FSB not explicitly mentioned in the "{terms}"
  deserve mention too -- e.g., I got the phrase "negotiating value
  of [bogus] patents" from Tilly,  Simo pointed me at useful stuff,
  Turnbull is mentioned above but there's no {turnbull} .....

   It's a work in progress.   I timed out on this revision.