Subject: Re: New angle on the patent problem
From: "Stephen J. Turnbull" <turnbull@sk.tsukuba.ac.jp>
Date: Wed, 15 Sep 1999 19:20:36 +0900 (JST)

>>>>> "shapj" == shapj  <shapj@us.ibm.com> writes:

    shapj> Rules that maintain a zero-sum game across national
    shapj> boundaries are generally a good thing.

BTW: "Trade is a zero-sum game"[1] means that war is the only way to
improve your lot.  I suppose you meant "level playing field"?

    shapj> If we were to succeed at getting them abandoned in the US,
    shapj> we would then be at a disadvantage in international trade.

Assuming that getting rid of software patents is a good thing on the
global level, most economists[2] disagree.  Most economists believe
that improving trade conditions and internal efficiency always
benefits you[3] even if you must do it unilaterally.

Is eliminating software patents a good thing?  And if so, how can we
convince policy makers and their economist advisors?

SOFTWARE PATENTS CONSIDERED HARMFUL

Patents are instrumental, according to the U.S. Constitution.  As
Bernard Lang pointed out, there are attractive[5] philosophical
reasons to suppose that there cannot be a natural property right in
ideas.  So under the current regime, we must consider what patents in
software would be intended to accomplish, what their side effects are,
the available "technology" for implementing them, and how well they
accomplish their intended purpose, considering the value of the side
effects as well.

I know you all know this, bear with me.  You may find it useful to
copy and pass along to the as yet unconvinced  ;-)

WHAT IS A PATENT?

The intended purpose is to "encourage invention for the benefit of
society."  Evidently it is assumed that
    o Invention is a good thing.  I will stipulate this.
    o Innovation is costly.  Note the change in the word.
    o Financial incentives are effective.  Again, stipulated.
    o Market-determined incentives are best.  Stipulated, but unlike
      the preceding two stipulations I expect there are people who
      disagree with this assumption, and I would be happy to discuss.

Effects (including the intended ones):
    o People who invent things are encouraged to innovate as well.
      (Invention = discovery; innovation means bringing a product to
      market.)  RMS[4] mentions a compression algorithm actually
      documented in a student paper before the application for a
      patent (eventually issued) on the algorithm.  _Getting inventors 
      like that student to publish and productize their work is one of 
      the express purposes of patents._ (intended)
    o People who don't plan on innovating personally are encouraged to
      publish and patent (well, usually in the opposite order) in
      hopes of collecting royalties._ (intended)
    o Some people are encouraged to conduct research with the
      intention of inventing and innovating._ (intended)
    o There are economies of scale in patenting; those who have a
      patent lawyer already will find it cheaper to patent a new
      invention.  Some bias toward oligopoly (but also a business
      opportunity that I will point out later).  (side effect)
    o There are economies of scale in licensing.  Oligopolistic bias.
      (side effect)
    o Simultaneous inventions are 100% out of the money, and their
      investments are "wasted."  (side effect)
    o An artificial[6] monopoly is created, which restricts the use of 
      the idea in order to drive up demand and price.  (intended;
      that's exactly how the system is supposed to work)

SIMPLIFYING THE ANALYSIS

As expensive as lawyers are, they're small change.  It's the
artificial restriction of quantity, and concomittent loss of user
benefits, due to (effective) enforcement of the patents that are the
significant costs here.  Getting patents issued, getting licenses,
etc, are small costs.  AFAIK, the only remedies available to a
patent-holder are cease-and-desist and possibly damages for past use
without a license.  I think it unlikely that an open source project
could be found liable for significant damages.  (It might be necessary
to incorporate, another small but annoying cost, but once you do if
you've got no revenues, there's not much the patent-holder can do
except force you to stop using the patented technology.  Let me know
if I'm wrong about that.)  If that's true, then there's little reason
to believe that patents should have much chilling effect on OSS
projects; they'll wait until the patent holder shows up, then invent
around it.

Another way to put this is to say that the "technology" of the patent
system is reasonably efficient, at least for the purposes of the
present analysis.  We've basically eliminated all the side effects
except the waste of simultaneous invention as significant factors.

I'm also going to assume that the patent inspectors get it right every 
time; pace Brian Bartholomew, I do not think that the patent office
gets it wrong 19 times out of 20, not by the rules it proclaims it is
following.  It's not a bad approximation to say they get things right
all the time, and it's not too hard to add in the costs.  Besides, I
get the right answer anyway.  :-)

WELL, ARE SOFTWARE PATENTS GOOD OR BAD?  TELL US!

OK.  So the question is "do the benefits of an increased rate of
investment in invention outweigh the costs of restricting usage of the
idea?"

Of course, pretty much everybody here agrees, more or less vehemently,
that the answer is "software patents are bad".  But why?  I think if
you ask about hardware, or drugs, most people here would more or less
reluctantly agree that patents make some sense.  What's different
about software?

As usual, the answer is to be found in the writings of that unsung
genius[7] of matters economic, Richard M. Stallman:

    Even an innovative program typically uses dozens of not-quite-new
    techniques and features, each of which might have been patented.[4]

Reversing the sense, _at the current threshold of "non-obvious"_
patentable (except for priority) software inventions occur many times
in _every_ program.  Furthermore, simultaneous invention is the
overwhelming _rule_, not the exception, in software.  It is the hidden
cost of suppressing these simultaneous inventions that is the
difference between hardware and software.  Ie, it is the hidden cost
of including in the patent-holder's monopoly people who, without the
patent, would not be considered to be "users of the patent-holder's
idea, directly or in derivative form".

In other words, the important difference is the frequency of
simultaneous invention.  (Sorry about that baroque syntax, I think it
parses.)  It is the significance of the side effect of wasting
simultaneous invention that distinguishes the cases of hardware and
software.

Of course, I don't have any numbers to support this thesis.  But it
seems rather plausible given the nature of programming, with its "joy
of always learning, which springs from the nonrepeating nature of the
task.  In one way or another, the problem is ever new, and its solver
learns something: sometimes practical, sometimes theoretical, and
sometimes both."[8]  Ie, programming is fun because it's invention
without the need to mess with grease for the gears, because it's pure
invention.  It stands to reason that in that situation the rate of
invention will be high, as will the rate of simultaneous invention.

I don't see any particular reason to suppose that the socially optimal 
level of "non-obvious" is "infinitely so"---ie, elimination of
software patents.  (I wouldn't be surprised to see that it was not
infinite, although I'm sure it's much higher than current standards.)

SO WHY IS UNILATERAL ELIMINATION A GOOD THING?

Because in the case of software, lack of a patent system actually
increases the flow of invention.  On top of that, by eliminating an
unnecessary artificial monopoly, it increases the flow of benefits to
a larger pool of users.  The cost of reusing software is significantly
higher than reinventing it in most cases, so reinvention (at the level
of the relatively clear-cut ideas that can be patented) is _not_ a
waste of resources.

A BUSINESS OPPORTUNITY

The one "problem" with this argument is the "business opportunity"
mentioned earlier.  That business opportunity is "patent search."
Today, it's conducted by skilled humans.  Expensive and time
consuming.  If someone finds a way to make patent search cheap,
_cheaper than inventing a solution yourself_, the entire argument
above gets turned on its head.

In other words, the patent system could be a crucial incentive for the
creation of that Holy Grail, reusable software modules, at a level of
abstraction far lower than anyone has ever really considered (AFAIK;
tell me if I'm wrong).[9]

OSS already does a lot of reuse, of course.  We have a few forks, here
and there, but the incentive for reuse is high.  We have cultural
divides (eg, Japanese programmers keep Japanizing programs developed
elsewhere, and the patches get rejected by the mainline cabals because
they're not really internationalization---so they get redone for each
upstream release), but there is serious effort being put into closing
those gaps too.

Question is, would the patent system with efficient search do this
more effectively?  (I think not, but that's a question for another
post.  Sketch of the argument for patents: since the costs of the
patent system are admittedly so high, the incentive to invent a better
search system is also high.  The same reasons to believe that software
invention is faster, cheaper, and "more simultaneous" than hardware
invention motivate the refutation of the argument that "if it were
possible, we'd already have it".  Further, the economics of royalty
payment would enlist patent-holders to enforce reuse much more
effectively than the moral suasion of cathedral-style project
leaders.)

Footnotes: 
[1]  Lester Thurow should be flogged for giving his book that title.

[2]  Those who agree with Jonathan are very smart and currently very
prominent.  But remember, they're prominent because it's good politics 
for say Bill Clinton to hobnob with advisors who recommend trade
policies that  protect existing oligopolies and their symbiotic trade
unions._  It is my opinion that the models they use are not robust and
do not really capture the real economic conditions in their structures.

[3]  In the sense of improving social welfare, accounting for benefits
both to consumers and to firms.  It is often true that oligo/
monopolistic firms and their symbiotic labor unions are harmed by good
policies whose benefits to consumers are nearly double-digit multiples 
of the costs to producers, and they yell a lot louder than consumers
who haven't enough expertise to know that their product could be 10%
cheaper if the workers and stockholders weren't paid 15% more than
their market value.

I would accept the argument that unilateral implementation of a good
policy may be politically impossible, so that we have to pin our hopes
on global implementation, in many cases.

[4]  But bogus, IMH and off the cuff O.  The same arguments apply to
creating rights to not have your air polluted, and then forcing
pollutors to buy them from you on a market AFAICT.  Works well where
it has been tried (admittedly, they picked the pollutants where it is
most likely to work well a priori).

[5]  http://www.gnu.org/philosophy/patent-reform-is-not-enough.html

[6]  In practice, the only real kind.  ;-)

[7]  I disagree with practically everything he says explicitly about
economics, but the analysis implicit in his writings about the
socioeconomic environment of software development is superb.

[8]  F. P. Brooks Jr., _The Mythical Man-Month_, 1982 ed., p. 7.

[9]  Of course, a compiler or interpreter is precisely a collection of 
such reusable modules and routines for automatically integrating them,
typically at an even lower level yet.  So VHLLs _could_ be considered
to be software reuse, but somehow I've never seen that point explictly 
made.

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
__________________________________________________________________________
__________________________________________________________________________
What are those two straight lines for?  "Free software rules."