Subject: Re: termless copyright and patents
From: Thomas Lord <lord@emf.net>
Date: Sun, 24 Sep 2006 11:57:46 -0700

I don't mean to blow off any specific criticism or question but
let me try to turn this around with a specific scenario and some
implementation possibilities.

* Setting: A Summary of the Imagined Law

  (Don't just leap on the details in this summary.   They
  are just background -- read them in the context of the
  scenario that follows.  *Then* you can come back on leap
  on the details of the summary. :-)

  A "limited patent" is a (an imagined) form of patent that can
  be obtained for as little as $75.  It is limited in term (say,
  to a maximum of 10 years, less if the filer so chooses).  It
  includes a mandatory offer for a public license, the price not
  to exceed $1M (less if the filer choses).

  Thrice the public license offer price is the
  cumulative maximum of damages that the holder of a limited
  patent can collect and, should such a sum be collected, a
  public license shall be imposed.  The originality and
  non-trivial nature of the claims of a limited patent must be
  obvious to domain experts considering factors such as historic
  and economic context, specificty of the work to a well-defined
  scope, documented effort in creating the invention, and common
  knowledge based on academic and trade publications and
  relevant public forums.  The scope of a limited patent shall
  be limited by the holder's practice including its practice as
  implementation and/or substantial and realistic public efforts
  to promote private or public licensing for specific practices
  -- that is to say that practices of a limited patent well
  outside those pursued or explicitly promoted by the patent
  holder are non-infringing.  With the exception of patents
  filed for prior to the date of this legislation's adoption, no
  practice of a limited patent infringes on another patent
  (limited or full) unless the holder of that other patent has
  pursued or explicitly promoted substantially similar
  practices UNLESS the party making that practice (a) was made
  aware of the possible infringement and (b) failed to give
  notice to the holder of the infringed patent.


* A Scenario Using Limited Patents

  Alice invents "A Method for Extending the Java Programming
  Language by the Addition of Hygenic Macros."

  It doesn't much matter for our discussion precisely what
  "Hygienic Macros" are.  Loosely speaking, if you've never
  heard of them before, they are a tiny bit like pre-processor
  macros in C ("#define MAX(A,B) ...") but they are both more
  powerful and more precise.  Many programming language experts
  have much good to say about them.  (Hygienic macros are an
  important feature of the Scheme programming language and you
  can learn more about them there.)

  Adding hygienic macros to Java is hard.  Alice needs to
  combine her deep theoretic understanding of hygienic macros
  with careful and complex extensions to Java syntax, clear
  ideas about how these macros interact with Java semantics,
  pragmatic clarity about how to implement the new features in
  typical Java compilers, and sensible treatments of macros in
  bytecode debugging information.

  There is prior art in this area: the "Java Syntactic Extender"
  http://people.csail.mit.edu/jrb/jse/index.htm

  Nevertheless, Alice takes a quite different approach, spending
  most of a year on the effort.  She pays $75 and registers
  limited patent of 7 years with a public license offer of
  $300,000.

  The Sun community process rules, we assume, were modified back
  in 2008 to permit language extension proposals covered by
  limited patents up to 7 years remaining and $500,000 for a
  public license.  Alice makes her proposal and renews a
  long-standing discussion about adding macros to Java.  A
  couple of proposals are put on the table, including Alice's.
  Following past patterns, some leaders (say, Gosling et al.)
  begin to draft a "union proposal" -- issuing drafts, accepting
  comments, and revising.  In just a couple of iterations, the
  union proposal looks an awful lot like Alice's and, in fact,
  is covered by several of her claims.  There are 6 years
  remaining on Alice's patent at this point.

  The formerly competing "Java Syntactic Extender" guys know a
  good thing when they see it and strike a deal with Alice.  In
  a frantic three months they put together a reference
  implementation of the proposal and distribute it under terms
  that forbid commercial use or copying and that require
  modifications to be returned (and reciprocally licensed) to
  them.  Their license does promise a reversion to open source
  in no more than 8 years.  They and Alice spend a few months
  cranking out little demonstration programs that illustrate how
  the new macro system will be a boon to many contemporary
  real-world problems faced by Java programmers.

  Alice and the JSE programmers put together what has become a
  somewhat standard-form public offer: they will sell commercial
  research licenses to the patent and code for $65K a pop and
  commercial distribution licenses for $100K each, with a
  promise that if they sell $500,000 worth, the code and patent
  will go public at that point.  MSFT buys a commercial
  distribution license and IBM a research license.

  A hacker in HateWTOLand registers a project on the FSF's
  Savannah host and begins a project to add the feature to gcj.
  Using boilerplate found on the web site of the non-profit
  "Software Labor Justice Center," Alice sends a polite cease
  and desist to the FSF who thank her for calling the matter to
  their attention, take down the project, and replace its home
  page with an essay about how patents make it impossible for
  programmers to work safely.

  Finally, Sun, RHAT, and Oracle make a collective offer through
  the Behlendorf Patent Liberation Front: each offers to donate
  $70K with a total of $200K earmarked by buy a public license
  for the patent and the reference implementation code (leaving
  $10K for the BPLF).  The offer is accepted and, once they are
  done splitting up the money, Alice has netted $300K and the
  hackers $65K.  That patent is liberated 5 years early.  The
  code is liberated 7 years early.  This is all one year after
  Java community consensus is reached.  $375K in transactions
  took place.  Alice put in about 1.25 years of labor for her
  $300K.  Each of the hackers cleared about $12K/month
  equivalent.  RHAT restarts the gcj project.  MSFT is quick to
  market with a very fast implementation.

  What's not to like?


* Discussion Questions

  Is any serious challenge to Alice's patent possible?  In other
  words, was the $75 basic filing fee reasonable?

  Did the limited patent succeed at creating an incentive which
  promoted progress in the useful arts and sciences?

  Was the hacker from HateWTOLand harmed?  If so, what would
  have been a better way for hygienic macros to be introduced to
  Java?

  Would the named vendors really behave as described in the
  scenario?   Would Alice?   The hackers?

  In this rosy scenario, the originality and non-trivial nature
  of Alice's invention is clear.   In what variations would this
  be less clear and how could it be handled?

  Who would have incentive to fund the Software Labor Justice
  Center?  Would the Behlendorf Patent Liberation Front be a
  welcome addition to the landscape?

  It would be impossible for the reference implementation in our
  examples to use GPL'ed code or code under any OSI-approved
  license.  Would it make sense to revise these licenses (or promote
  new licenses) that would relax this constraint?


* Faking It?

  Limited patents don't exist.  It isn't immediately clear how
  to write legislation that would implement them or how that
  legislation would interact with international patent treaties.

  Does the scenario above suggest a patent regime that will
  appeal to those who have an interest in open source software
  and open standards?

  If it does, would it be worth the investment to "fake it?"
  By "fake it," I mean implementing something like the above
  scenario, but without legislation.   Namely, the "Try Out
  Limited Patents Foundation," an NPO, might play the role of a
  future PTO.   Alice would pay the foundation her $75 dollars.
  They would subsidize and additional several thousand to apply
  for a normal, full-protection patent while implementing
  Alice's limited term and public offers through a series of,
  well, irrevocable public offers.   To the $375K spent on
  Alice's and the hackers' work, a "Try Out Limited Patents
  Foundation" would, unfortunately, add some several $10K of
  additional overhead.   They would have to manage some risks
  that would not be present in a legislated solution.
  Nevertheless, such a demonstration might lead the way to a
  clear legislative formulation and optimization of the idea,
  along with an empirical test of whether or not it does, in
  fact, promote progress in a palatable way.

  Faking it would be unable to implement one part of the
  proposed legislation:

      With the exception of patents filed for prior to the date
      of this legislation's adoption, no practice of a limited
      patent infringes on another patent (limited or full)
      unless the holder of that other patent has pursued or
      explicitly promoted substantially similar practices UNLESS
      the party making that practice (a) was made aware of the
      possible infringement and (b) failed to give notice to the
      holder of the infringed patent.



-t