Subject: Re: GIF/LZW patent
From: <stephen@xemacs.org>
Date: Thu, 28 Sep 2006 15:15:10 +0900

Thomas Lord writes:

 > You miss mine, then.   Copyright creates fiction around the whole work
 > which is most certainly innovative in the most important sense, at least
 > in an example like SCM.

I disagree.  All of the large programs I'm even a little familiar with
(which includes three Lisp engines) are modular.  I can't imagine SCM
is any different.  So we can start reducing.

 > Copyrights protect complex texts in which the value resides in their
 > complexity and intricacy.   Change a few words of Moby Dick and
 > you still have a great novel (if yr into that kind of thing) even though
 > any big easy-to-sum-up-in-a-page-or-two innovations are extremely
 > diffuse in that work.

Sure, ergo copyright.  

 > Most interesting programs are like that.

Nonsense.  As I pointed out, you wrote up a claim for an interesting
easy-to-sum-up-in-way-less-than-a-page innovation.  I'm sure there are
many, many others in SCM.  Each one can be patented, in principle.
The effort to do so puts a brake on it, though.

I'm not denying that there are interesting aspects to the whole.  I
am, however, saying that it may be better to incent innovation in ways
that get the innovation described explicitly.  For example, if there
are novel interactions between modules of SCM, they can be described
in patent claims of their own.  It just isn't clear how much you lose
by requiring that the "gestalt" get abstracted a lot before granting
IP in it.

 > > The real problem with Larry's argument is that to make the friction of
 > > copyright go away, you just close your eyes.  Ignorance *is* an
 > > excuse.  With patent, however, ignorance is no excuse.
 > 
 > Sorry but no.   For most interesting programs, ignorance rewards itself
 > with inefficiency.   "It is easier to modify an existing program than
 > to write one from scratch" -- RMS (paraphrased, personal communication).

Of course.  Nonetheless, ignorance is an exploitable tradeoff when
faced with copyright.

Furthermore, *if* reinvention is an inherent part of the software
development process as Jamie and Forrest describe, then such ignorance
is also inherent, and helps lubricate the process.  We shouldn't throw
sand in the gears.

 > Hence why I have been insisting on tying scope to practice where
 > "practice" includes specific promotion.

Unpack, please.

 > >  > But there's no way to formulate a useful claim.
 > >
 > > Looks to me like you just did.
 >
 > To be clear, I've barely touched the surface of SCM.  You should
 > understand what I said about SCM as an excerpt from a much, much
 > longer lecture.

You said "a", so I pointed out *one*.  One point is that writing up
all of them is going to be expensive.  So they'll choose a subset,
some of which may be very low-level, others high-level.

If SCM is open source, it becomes prior art for all the rest.

(Assuming the existence of patents) what more could we want?

 > SCM gets to monopolize a particular text which conveniently
 > embodies all of those choices -- not to monopolize the interesting
 > set of choices.

You're contradicting your Stallman quote.  There's no legal monopoly.
But economically, there is friction, even a barrier, to entry, and
that is what is meant by monopoly.

 > > With copyright, sure you can say "we want to license from line 528 to
 > > 10016" but this is going to be much more complex to negotiate.
 > 
 > I dunno.   Seems to work ok.    Why did RHAT by netscape code again?
 > Surely not for that "huge" customer base, right?

Yes, you can negotiate for the whole thing.  But it's hard to break up
into parts.  Patents, on the other hand, naturally come piecemeal.