Re: [tlug] RE: Shuttleworth Interview at Red Herring (Roger Markus)

["Stephen J. Turnbull" <stephen@example.com>]

Fergal Daly writes:

 > By "enforceable" I mean will not be invalidated by the first court
 > that looks at it. There are plenty of patents that have been granted
 > and then invalidated when they were challenged.

Sure, but that requires a lot of work, demonstrating prior art or
other technical deficiencies in the application.  In the US at least,
you'd have to contest those patents one by one.  The fact that one
software patent was invalidated would not make a difference to the
next one; you'd still need to come up with prior art or technical
deficiencies in the application.

 > >  > I don't think the court that set the precedent could possibly have
 > >  > known or understood the consequences of their actions and since
 > >  > subsequent courts follow that precedent, they are not analysing the
 > >  > overall economic consequences of software patents as a concept.
 > >
 > > Who needs to?  The analysis is trivial: a monopoly is granted, which
 > > results in restrictions on use and transfer of income to the patentee.
 > > The latter is the purpose of a patent, as explicitly stated in the
 > > U.S. Constitution; the former is an adverse side effect accepted for
 > > the sake of the encouragement of the useful arts.  I'm sure that the
 > > courts that hear such cases understand this---they mention it often
 > > enough in their decisions.
 > 
 > The law is supposed to promote the development of the useful arts.

That's right:  the *law* is supposed to do so.  *Individual patents*
do not bear that burden.  Furthermore, the judgment as to whether the
law promotes the useful arts is generally a *legislative* judgment.
The courts will take action to overturn such a law or modify the
principles of enforcement only if they judge that constitutional
rights have been violated.

The court can't generally change the applicability of a law to certain
domains; it can only decide whether the language applies to a domain
or not.  Patent law as a whole (hardware, software, genetic mapping,
chemical formula of pharmaceuticals) is what's in question.  The court
simply decided that discrimination against software was not justified
by the language in the law.

Another way to put it is that there is no such thing as a software
patent in U.S. law.  There are only patents.

 > > Precisely, if you restrict "consequences" to legal ones.  The
 > > legislature may have screwed up social or economic policy when it
 > > wrote 35 USC.  That is not the court's problem, and it would be
 > > severely criticized, perhaps even impeached, if it tried to rectify
 > > what it considered to be a mistake due to overbroad language.  That
 > > would be changing the law ex post.
 > 
 > Couldn't the supreme court could declare the law unconstitutional if
 > it was shown to stifle development.

No.  That's the legislature's job, not the court's.

 > This is what Lawrence Lessig tried to do with the ever-growing
 > copyright terms. He failed but he was given a hearing

Unlikely.  The court doesn't give hearing to law professors.  It hears
cases.

I assume that what you are referring to is an amicus brief in the
Eldridge case.  There was a similar one submitted by a group of
economists (the first time I've ever seen Ken Arrow, James Buchanan,
and Milton Friedman agree on a policy question!) which tried to argue
that extensions were economically unjustifiable, as they can't create
incentive to produce a work that already exists.  The court decided
AIUI that its hands were tied; the law was constitutional, as there
are no criteria to judge what a "limited time" is except those that
Congress itself decides.  They were pretty scathing about the
retroactive extensions IIRC, but again they decided that Congress
wasn't bound by any Constitutional argument not to extend.

 > The EU commission's push towards SW pats is not based on a serious
 > economic study. I believe you are referring to this document (1st
 > link)
 > 
 > http://www.google.com/search?q=%22The%20Economic%20Impact%20of%20Patentability%20of%20Computer%20Programs%22

One of them, yes.  I don't have the others offhand, and they're
currently buried in the detritus of an office move.

 > Right, so there's an example of license problems preventing reuse but
 > you seem to be saying that somehow more proprietary licensing would
 > make for more reuse.  Yes it sucks that every function on my computer
 > can be done with KThis and GThat and they have slightly different
 > features etc but within the K-world and the G-world there is a lot of
 > reuse.

And there is one hell of a lot more reuse in the M-world, since you
either reuse M-libraries or your program doesn't run.

 > > It's also unobvious to me what it means to talk about reuse of glibc,
 > > which implements the ISO C and POSIX library standards, of which there
 > > are probably a dozen or so major implementations.  Is compiling code
 > > originally targeted for FreeBSD on Linux and discovering it Just Works
 > > reuse of glibc?  I don't know!
 > 
 > glibc is a slightly different kind of reuse. Those other dozens of
 > implementations of the standards all run on their own platform. Sun
 > has one, HP has one, SGI has one, MS has one. glibs works on many
 > platforms. If I write a new OS, I can write my own libc from scratch
 > or I can reuse glibc. I don't know what the licenses are like but I
 > doubt there's any way for me to reuse any of those proprietary libcs.

There are *several* free ones.

Nor is writing an OS the most common way to reuse the code in a libc.
At the level you seem to be talking about, libc is not used by the OS,
it's simply distributed *with* the OS.  That's reuse as a product, but
not use as code!  By far the most important way to reuse libc as code
is to write programs that call it.

 > Sticking with more usual forms of reuse, there's still ncurses,
 > berkeleydb, readline, large chunks of CPAN, rubyforge, python-thingy
 > (I forgot the name), libboost. Check how many packages on your OS
 > start with "lib". Some are used by just one app but many of them
 > aren't. It's just not true that "the only time software developers
 > actually reuse software is when it's proprietary",

That was an exaggeration for effect.

 > and I got 603 lines. 174 are only used once but there is also
 > 
 >     557 libz.so
 >     360 libpng12.so
 >     358 libexpat.so
 >     338 libjpeg.so
 > ...
 >     164 libgcrypt.so

But suppose that all 338 of the programs that use libjpeg.so are image
display applications a la xv and ImageMagick?  That would indicated
massive duplicative effort.

The statistics you are giving simply don't bear directly on the
question of how to increase reuse or if it is possible, which is still
wide open.

 > > Of course, in some sense.  But that's not the question.  The question
 > > I'm asking is would there be even more reuse in *socially beneficial*
 > > ways if those libraries were proprietary in some way.  The answer
 > > without the qualifier is pretty obviously no; but the answer with the
 > > qualifier is very controversial
 > 
 > I don't understand what "socially beneficial reuse" means. How is it
 > different from other types of reuse?

If 338 programmers all write their own versions of ImageMagick just
because they can, 337 of them are wasting their time.  That's not
socially beneficial.

I don't claim to be able to measure this.  What I do know is that the
people most opposed to various kinds of IP invariably advocate the
freedom to hack rather than production of software that the users
value at the lowest cost.

 > >  > You'll have to elaborate on that point.
 > >
 > > I already did, as you quote:
 > >
 > >  > > Note that Stallman has been badmouthing my project for more than a
 > >  > > decade, and has *never* reused anything we've developed since 1994.
 > 
 > I don't understand the reasons for this non-reuse (maybe not wanting
 > to contribute to something that hasn't been assigned to the FSF, maybe
 > just being RMS). I still don't know how this is symptomatic of
 > non-proprietary software or would be improved by more proprietariness.

They're not relevant.  The point is that free software is not designed
to increase reuse, it's designed to allow programmers to rewrite
existing software, and one particularly important example clearly
doesn't give a damn about reuse.

 > I'm thinking of the betamax case for example
 > (http://en.wikipedia.org/wiki/Sony_Corp._v._Universal_City_Studios)
 > where the new circumstances were the existence of video recorders.

What was *reversed* in that case?  AFAICS, the court simply said
"these are different, here's how to deal with the differences without
screwing up existing precendent".

 > I don't know about US Supreme Court but a colleague mentioned in the
 > UK a few years ago, the House of Lords (the last step on the law
 > ladder in the UK) specifically declared that it was not bound by it's
 > previous decisions. That is, it could change its mind in the future.
 > >From what you say, it sounds like the Supreme Court can't do that,

AFAIK there's not a limit to what it *can* do; I think it's a pretty
safe bet that it *won't* reverse itself unless it finds a technical
problem.  The Lords are a different case; they are not legal
specialists, and it seems likely that they will make mistakes that
they would want to reverse.  SCOTUS is a bunch of legal specialists,
and they're damn good at what they do.  They too will make mistakes,
but the prevailing opinion is that it is better to maintain precedent
for the sake of the vast majority of cases where the decision is quite
good, and work around (eg, via superseding legislation) the small
number of bad decisions.