Re: Does the LGPL encourage sharing of code?

Anthony Taylor wrote:
> 
> More of my verbosity.  Lots of good stuff from Roger taken out; this is
> the Reader's Digest Condensed version.

Gee, I feel like milk :)

> Roger Fujii wrote:

> >well, only if you think GPL is 'free' speech.  Personally, anything that tries to
> >exclude other valid speech is hardly 'free' IMHO.
> 
> It doesn't try to exclude other free speech; it tries to prevent
> commercial plagiarism.  This is no different from me writing a dandy
> story and publishing it on the web and saying, "You can use this, and
> add to it and extend it, but you can never include my parts in anything
> you sell; and you must also allow any derivitives to be distributed
> under the same terms."

While I might let this go for LGPL, it certainly isn't true for GPL.
To show this, just think of a GPLed library.  If you use it, you aren't
extending it, nor are you including in a shipping product (assuming that
it is part of the system).  BUT you are *still* required to GPL your work.
It is pretty much exclusionary of anything that isn't GPL.  If you want
to prevent plagiarism, use MPL, or AFPL.  You should view GPL like a
benevolent dictator.  It might do some good, but conceptually, it's still
a bad thing.

> To say the GPL steals is ignoring the fact that you *still* maintain
> copyright over your own code.  (This offer null and void if you assign
> your rights to the FSF.  But using the GPL does not automatically do
> that.  You have to fill out forms and everything.  Not that *I'd* do that.)

Mostly true, but there's still an inaccuracy.  Unlike the 'english-challenged'
that claims that lindows/transgaming 'stole' code, even though it was freely
given to them to do as they wish (more or less), GPL lays claim to things
that are NOT given away.  Take a plugin to gimp.  Even if I use *no* GPL source
code (other than header files - it's arguable if that should count or not), if
I write a plugin for gimp, it *must* be GPLed (or at least, the GPL will make a
claim on it).  So, in this case, GPL will try to 'take' my code.  Stealing something
doesn't only mean you don't have access to something.  It is that you lost CONTROL
over that something.  Using the logic you gave, you can *never* steal an idea, and
this simply isn't so.

> The GPL does one simple thing: It keeps GPLd code out of proprietary
> products.

and visa versa.
 
> >Anthony Taylor <[email protected]> wrote:
> >
> >>!!!! WARNING !!!!
> >>Long, boring message follows.
> >
> >wan't THAT long :)
> >
> Ahh.... but *BORING!*  And I don't get any better.

try taking DiffEq at 9AM....
 
> >>!!!! Warning !!!!
> >>Roger Fujii wrote:
> >
> >HEY! :)
> 
> Oops!  Sorry.  Now I'm a bit shame-faced.  Didn't mean it like that.....
> The warning was for the stuff *I* wrote.  Really.  I mean it.  :)

I know... I found it quite humorous.  Especially after the "boring" statement :)
 
> >Actually, you can't look at it like this.  A simple example (in real life)
> >would be teachers (at least in the US).  Their value as a whole is high,
> >but their "price" is low.  Or even more simply, air is important to all of
> >us, but it has no (commerical) value.  While there is usually a relationship
> >between import and value, there doesn't have to be.
> 
> Yes.  That's my point.  The economic arguments against the GPL have all
> been, "It doesn't let you make money off the code, because you can't use
> the code as a basis for a proprietary product."  This is invariably
> followed with, "Ergo, GPL will destroy the software economy."  This is a
> poor basis for an economic analysis, and ties economic contribution with
> selling price.

Actually, this isn't quite right.  The economic argument always has been 
that as it competes with a "free" (as in beer) price, and that you can't
leverage off of it (as that would force you to GPL your work), it puts a
very large pressure on any company competing in that space.  As the current
"software economy" works on a price/unit model, I don't see how you can
say that it isn't going to destroy the industry.  Now, you may claim that you don't
like
the price/unit model, but if you don't, please explain what model with >some<
evidence of long term success would work with the GPL.
 
> > Using Ford's weren't exclusinary.  If you use a model T to haul goods, you don't
> > have to give it up.
> 
> And if I use a GPLd word processor, I still own the document. 

questionable if you use a GPLed macro.  If you use a GPLed GUI editor, your GUI
has to be GPLed.  Going back to the Model T example, I probably should have
used a "trailer" than "haul".

> Using GPLd software is not exclusionary.  You can use GPLd software on
> proprietary systems, and proprietary software on GPLd systems.  You just
> can't co-mingle the source code.

This is NOT true.  Think back to the GPL library example.  And it *is*
exclusionary, as it *only* works with itself.  Why else would the phrase
"GPL compatible" come from?  Things that are "GPL compatible" means that
it can be turned into GPL.

> Secondly, the economies of scale talk about production efficiency.  It
> is cheaper per-unit to produce 1000 of a particular unit than to produce
> a single unit.  In the software world, it is exactly the same price to
> produce a single unit as it is a billion units.  So I don't see how the
> GPL affects the economy of scale.

If you want to get technical, the correct term here is that "the average
fixed cost" doesn't go down much in a GPLed world as the copies goes up. 
So, using your example, I have to sell the product at $1B for the first
guy, instead of selling it @ $1 to a billion people.  The point here
is that "economies of scale" implies that the cost/unit goes down as
number of units goes up.  The GPL doesn't follow this, as you can count
on any income after the first copy.
 
> As far as software services go: there is always work maintaining and
> extending systems, in systems integration, custom programming
> (especially for web services and databases), and systems support.  These
> industries are not tied to particular software; in fact, the more stable
> the software, the better it is.  And these industries are growing quickly.

SHOW ME.   Yes, there is growth, but look at this from linux weekly news (hardly
a GPL detractor) http://lwn.net/2002/0321/  
I see no evidence of any great successes on what you just said.
 
> What you are talking about is re-use of code in proprietary products.

There is a difference between re-use and "be able to use".  As code is in modules
by CALLING them, you can USE code, without COPYING the code. 

> (And in BSD/X11 code, too, which I admit is a major problem.  Perhaps a
> fatal one, for me.)  Since this is all covered by copyright law, let me
> put it in terms of copyright:
> 
> Using GPLd code in proprietary products is stealing, just like using
> copyrighted fiction in your own novels is stealing.  This is called
> "plagiarism."  It doesn't matter if the fiction is available on the web
> for free, or sold in the most exclusive quarterly, this is still
> plagiarism, and it is wrong.

while this is true for copyleft,  It is just a subset of what the GPL is.
Using your example, if I write a book and sell it WITH your book (doesn't even
have to reference it.  Let's say it's Roger's favorite stories collection).  My
book must be GPLed.  This is (IMHO) *just as wrong* as plagiarism.
 
> Anyway, the limited use rights given by the GPL are far more liberal
> than the use rights given by proprietary software vendors (that is, none
> at all).

GPL isn't high and mighty.  Sun gave code back to the BSD project before GPL even
existed.  It is *A* license, and it is mostly responsible for popularizing
open-source,
but it doesn't have a sole claim to virtue.

> Since we're talking about code freedom, defending the biggest
> offenders of code freedom while damning the second biggest offender
> seems skewed, somehow.

huh?  missed something.  I am certainly not defending M$.  Not sure who
you are talking about here.
 
> >but unless you can make the mapping between your 'worth' and value, you are
> >not talking anything concrete - you might as well talk about religion.
> Just because you can't quantize a value doesn't mean it has no value. 

no. But it makes the discussion difficult.  And I didn't say it had "no value"
(unless you are equating "religion" with "no value").  

> And finding out how much people will pay is hardly adding to the value.  And you
> *can* quantize value; there are always people willing to say that the industry has
> lost billions due to the Melissa virus, or some twaddle such as that.

I was only saying that you can do a comparison if you can quantize the value.  If 
you can't, you are in the "apples to oranges" comparison, which is useless.

> Can you map the value of cheap transportation to a direct monetary contribution to
> the economy?  No.  It's too complex.

I can try to do the mapping.  It might not be RIGHT, but at least in doing so, I
enumerate
how I come up with the mapping. 

> Economics is not a well-developed science; it's more in the "fire/air/water/earth 
> chemistry" stage.  There are a few obtuse cause-and-effects figured out, but there is
> no way to map true economic contributions of a product to the economy as a whole.

Just because you can't definite it precisely, doesn't mean you can't come up with
accurate statements.  "GPL -> software" as "anarchy -> government".  I don't think
you can find anyone who thinks anarchy on a large scale is a good thing.  
 
> Basically, if software contributes to the economy in general, it will do
> so whether someone charges for it or not.  And if it adds average value
> V to the economy for one person, it makes sense that it will add
> approximately value N*V to the economy for N people using it (assuming
> those people all receive more-or-less the same value).  So the greater
> number of people using it, the greater value it will add to the economy.

But this ASSUMES that you are not REDUCING the value that the N people put out.
Remember that GPL requires to distribute your stuff for free.  In GCC's case,
it's more complex, because the GPL in this case reduces diversity.  How one
values diversity is pretty arbitrary. 

> The mapping is already done.  We have determined that software
> contributes to the economy, though  think we may overestimate that
> contribution.  We have further determined that having a store of
> available source code adds value to the economy.  (This was established
> when you said some companies would need the source code as a basis for
> their product.)  Does it not make sense that the more code that is
> available, and the more software that is available, the greater the
> contribution to the economy?  Or is there some limit of which I'm not aware?

This is true for X11.  With GPL, which requires derivatives to be free, it can
easily be argued that it DECREASES the economy.  After all, if you can only make $30,
instead of $300, the net result is that you make less of a contribution.  You might
say that there is others that MAKE more money than you lost, but I haven't seen any
evidence of this.  No linux distribution I know of is profitable.  This isn't a good
sign.
 
> >>And: economics isn't the only reason to do something.  Why should there
> >>be an economic argument for the GPL, when the GPL isn't about economics?
> >
> >It *IS* about economics, as it takes a stance on intellectual property.
> >The whole notion of property is a pretty fundamental economic principle.
> 
> And the core philosophy of the GPL is that ideas and knowledge are not
> property, and the greater the distribution of knowledge and ideas, the
> better off the world is.  The GPL attempts to remove GPLd code from the
> world of economics.

Huh?  If you compete with economic entities, you are not removed from the world
of economics.  Taking what you said, then you (or whoever that is following
this core philosophy) is against *all* copyrights and *all* patents.  Realistically,
it is quite like communism in its idea of property.  The fallacy of both is that
it fails to recognize that anything that can think responds to rewards wrt their
behavior - it's pretty fundamental.  

> It is an imperfect solution to a very large problem: the ownership of
> thought, knowledge, and ideas.  The only reason to espouse the GPL is if
> you feel the hiding of knowledge for profit is a bad thing.  I
> personally believe the hiding of knowledge for profit is unethical for
> the same reasons Brett has said the GPL is unethical: because it hurts
> *way* more people than it helps.

You seem to ignore that the GPL thinks all things that can't be converted
to GPL is proprietary.  So, if I do something like AFPL, or SCSL, where I show what
I do, but don't allow you to redistribute it (which gets rid of your *and my*
objections about hiding work, but still preserves property), it is still considered
"incompatible" by the GPL.  So, "showing work" is not what GPL is ultimately trying
to do. 

> Now: I know that copyright covers works, not ideas.  But most of the
> reasons to hide code are to hide the knowledge contained within.
>  Imagine if we'd had view-only access to all the code for all the
> software in the world; it would have made Microsoft's current position
> impossible, and would have removed the very core reason why RMS started
> the FSF and GNU project in the first place.

Nope.  RMS would have done it anyway, as what he wants is unfettered access and
use of everyone's code.  The core reason is that rms believes that you should 
not be ABLE to withhold code for others to take.

> Science and technology have advanced so quickly because of the free flow
> of thoughts and ideas.  If we expect to advance in the software field,
> we must follow those ideals; otherwise, we are dooming ourselves to
> MS-Windows-level quality for the rest of our lives.  And as both NeXT
> and BSD have proved, we should be far beyond that already.

But as BSD (and even linux) isn't commodity (compared to the evil oses),
the free flow of info isn't what's causing this condition.  The problem
with GPL is that as it is anti-proprietary, it can't pick and choose from
the best (as some proprietary stuff IS the best).  My hope is the BSD
will overtake linux in a couple of years due to this.
 
> >It's an imperfect world.  There will always be people who break the rules.
> >The point is to make it such that the rules are good for the majority that
> >follows them.
> >
> Exactly!  I think we just disagree on which rules those are.... :)

or more precisely, the EFFECT of the rules.  You can't make these rules
in a vacuum.  You have to consider how everything else fits together.  Sort
of like playing jenga... :)
 
> >But for me, I would face either having to GPL the program, or stop selling it.
> >If you want reciprocation, look at MPL, which does it correctly, AFAIK.
> >If I change what YOU wrote, I see no problem in giving you changes.  However,
> >if I just USE (by calling) what you wrote, without using what you wrote, I
> >have great problems with you making claims on something that you didn't write.
> 
> There is still the LGPLd version, which you originally compiled against
> (once removed).  You couldn't use the newer, GPLd version, but you could
> still use the older version.  You would still have every right to sell
> your software, distributed with the old LGPL library (and make the LGPL
> source code available, of course).

and make it linkable (and go through all the hoops that you need to go through
to distribute the LGPLed library), and deal with any incompatibilities that might
result
(interprocess communication with the system installed libraries might break).
Overall, a real PITA.  Effectively, you have been screwed.  It's also arguable that
if you ship such a combined work (your work plus the LGPLed library), it CAN
be claimed that your code is a "work as a whole" and thus your work must be
LGPLed. 

> >>Why is it all right for a company to do this, but not an individual?
> >
> >I don't know of any modern license that makes the intellectual claim that the GPL
> >does.  Besides, I never claimed it was all right for them to do it either.
>
> If you contribute to GPLd code, and you distrubute that finished
> product, the only way you can avoid including the GPL in a proprietary
> product is by also licensing your code under the GPL.  This does *not*
> remove your ownership of the code you wrote; you may use that code in
> any way you see fit.  But to distribute it co-mingled with GPLd code,

It doesn't have to be co-mingled.  There only has to be a path, no matter
how indirect. 

> you must also license it under the GPL.  Let me restate: The GPL does
> *not* exert ownership of your code.  The GPL dictates the terms under
> which GPLd code may be distributed, and that is: without proprietary code.

I never said you lose ownership..  What I have said in the past is that
you licensed it to the FSF (in a more animated discussion).  Here, I only
said that it makes a claim.  
 
> This seems perfectly equitable to me.

You keep only saying half of what it does.  GPL *also* says that if you USE
("derived", as GPL calls it) it, you must be GPLed also (with some explicit
exceptions).   If GPL restricted itself to what you have said, it wouldn't
be objectionable.  The problem is that as it *is* anti-proprietary, it claims
a whole lot more. 
 
> >
> >of course, if you want portability, use java :-)
> Damned straight.  :)

FreeBSD ships the j2sdk.  Something no GPLed distribution can ever do.....
 
> >>There was a time when Linus said it would never run on anything
> >>but x86.  No, NetBSD's superiority comes from better, more-modular code.
> >
> >and maybe its license? :)
> 
> Okay, okay!  Its license, too.  :)

heh. :) :) :)
 
> Honestly, I do think the BSD is a more-honorable license.  The GPL is
> rather cynical at its core.  Granted, I too am cynical, but I appreciate
> the purer nature of the BSD/X11 licenses.

I don't see it as being cynical.  It has an agenda, and it's pretty good
at heading towards that goal.  If you believe in the agenda, you should
use GPL.  If you don't, you should choose something that you're comfortable
with, but choose something else.  The main problem in this license discussion
is that people decouple the agenda with the license with the GPL, and I don't
think in the big picture, you can do that. 
 
> >>I don't claim the GPL is better than the X11 or BSD license; it has
> >>different goals.  Brett claims those goals are to destroy the software
> >>industry;
> >
> >But this is a TRUE statement.  It is/was a stated goal of the GPL to
> >eliminate propriatary software.
> 
> We are saying two different things.  I say "software industry;" you say
> "proprietary software."   GPL was written to keep code out of
> proprietary products because the author of the GPL believes proprietary
> software ("keeping secrets") is morally wrong.  He also doesn't believe
> the end of proprietary software will be the end of the software industry.

I am using the phrase just to be "correct".  Remember that GPL considers
anything it can turn to GPL as being proprietary.  It makes no distinction
between AFPL and an M$ EULA.   As for ending the industry, I should
point out that you can destroy something without ending it.   I don't think
the statements are contradictory at all.  I don't believe the overall
state of computing will be BETTER (quality-wise) for the user.  It'll
be cheaper, that's for sure.

> Me, I think there's no way we can tell.  I do think keeping knowledge
> and ideas secret for a profit motive is wrong.

So, coca-cola is wrong?  Why have companies?  Let's outlaw corporations,
political parties and all other occupations with competition for some
profit goal while we are at it?

> But I also think it's
> wrong to take without giving (this works against proprietary software
> using free code in all cases, and GPL using free X11/BSD code).

Definitely hypocritical.....
 
> >>The original claim was that it displaced better compilers because of its
> >>license.  I claim it displaced more efficient compilers because it is
> >>portable.  Really, that's all I was driving at.
> >
> >Not sure if I would give you that one, as most software companies don't do
> >multi-architecture projects.   That fact GCC runs on an ARM chip wouldn't cause
> >most people to use GCC.  Now, COST, on the other hand.....
> 
> Where is GCC used, mostly?  How many proprietary products are
> GCC-compiled?  I'd have to see more analysis before I'd commit one way
> or the other.  My *impression* is that most proprietary software is
> based either on MS products (for the Wintel world) or on the compilers
> of the manufacturers (from the Big Iron and Small Iron worlds).  But, I
> could be completely wrong on that.

depends on the platform and what you are trying to compile.  I know
that compiling stuff like gnome practically REQUIRES the gcc - a trend
that is all too common nowadays.  Wine won't compile on anything but the gcc. 

> Anyway, this is *so* way OT.

Not like this is in wine-devel, nor is there a lot of traffic here (other
than the self-generated ones :) )

> Not that I mind; I am invigorated by these sorts of discussions. 

I'd be too if it weren't for the time....

> Not that I think we will change each other's
> minds, but we will at least come away with a better understanding of the
> other's side.  And that is an important goal.

My only goal is to get people to do the right thing for the right reasons.  
I just ask for some consistancy (yeah, yeah... call me a hobgoblin :) )

-r