re: re: freedom

["Robert G. Brown" <rgb at phy dot duke dot edu>]

On Wed, 2 Apr 2003, Gangolf Jobb wrote:

> 
> >I am puzzled here. Why do you need a free library to develop a 
> >comercial software? People of GSL (and other GNU projects) simply say 
> >"we give you this software for free provided anything you make with it 
> >must be also free". Seems like a fair deal to me. 
> 
> hey you copyleft-communist,

I >>was<< enjoying this well enough, but if the conversation turns to
flames I'm going to change my mind as will most people on the list
(where this whole conversation is close to noise anyway -- so it needs
to be polite and enjoyable noise).  Having said that, I might as well
respond, hopefully in an enjoyable and non-flaming way.

> don't you think it is a too simple and stupid view to believe that all 
> software is equal? one must clearly distinguish between software whose 
> development has been financed from public or other peoples money and 
> software that somebody writes on his own risk without being paid. 
> as far as i can see do most people contributig to a numerical library 
> like the gsl work in universities or some kind of governmental
> institution, and all governmental institutions that i know including the 
> universities are explicitly supporting the commercial use of their 
> scientific results. people working there do simply do not have the right 
> to prohibit the commercial use of their work by putting it under the gpl, 
> because commercial use is in the public interest. unfortunately, people 
> deciding about fundings are not generally aware of this and someone should 
> tell them. 

The author of any work, anywhere, at a University or elsewhere, who is
not doing "work for hire" (and publishable work done at a University or
on a grant is generally NOT considered work for hire unless the grant or
terms of employment explicitly spell that out) retains copyright
"automatically" in the eyes of the law.

If I write something, I can publish it or not as I choose (and I can
find a publication venue willing to do so).  In that publication
process, I (as author) get to pretty much absolutely control whether and
how copies may be made or sold.  That's what the word "copyright" means.
I have the right to regulate copies of my work and sell them or
distribute them for free as I wish.

As the copyright holder, I can pretty much set my own terms for the
"license" to make, own, resell a copy of my work.  If I want someone to
rub blue mud in their belly and stand on their head as a contractual
condition for owning a copy, I imagine that this would stand up in court
if I were ever insane enough to try to enforce it.  It would be >>easy<<
to require people to send me money, or pay me royalties, as a condition
of making a copy for themselves or for resale respectively, as this is
in fact part of what the law protects -- my right to make money out of
something I created and presumptively "own".

Suppose I wrote a poem (in fact, I've written many of them, all
distributed under a modified open publication license, as one can see on
my website).  I retain copyright automatically, but the legal process is
helped if one sticks words like "Copyright Robert G. Brown April 2003"
somewhere on the document or collection of documents so I generally do
so.  

I'm a nice guy, so I let you read my poem online for free (de facto
granting you the "right" to own a temporary copy of my poem inside your
browser or other web engine).  Hey, I'm a nicer guy.  I include with my
poetry collections a license that gives you certain VERY SPECIFIC rights
to make further copies subject to certain conditions.  You can read them
online all you like.  You can make paper or electronic copies so you can
read them again when you aren't online.  You can even make copies and
bind them in leather and sell them to people, as long as you don't make
a dime of profit from doing so, including the backhanded way of owning a
chunk of the publishing company that does the binding.

However, if you take >>my<< poem, publish it in a book bound in leather,
and sell the book for all sorts of profit without dickering out a fair
share with me and signing contracts and so forth to regulate the
commercial transaction, I'll have to beat you to death with your own
torn off leg, in a legal and metaphorical sort of way.  That's not
communism, it is pure ownership -- they are MY poems and >>I<< get to
regulate who can own copies and how they can be distributed.

Note that you have NO RIGHT to go ahead and publish them for profitable
sale without FIRST getting my permission to do so, REGARDLESS of the
conditions I impose in order to grant my permission.  They are my poems.
You can't sell them, and can only possess them at all to even look them
over at my whim, until roughly the year 2090 or so, if they don't
lengthen the term of copyright yet again.  

Perhaps I'll grant you permission to make a single for-sale-for-profit
copy if you pay me $100 billion dollars.  You may be unlikely to pay
that, but you can't go to court and "insist" that this is unfair because
you can only sell that copy for fifty cents -- the court has no legal
basis to compel me to provide you with a license to republish and sell
>>at all<< until we sign an agreement.  You can only choose to pay or
not to pay, to publish with my permission and license or not publish (or
publish without my permission and risk my righteous wrath with the full
weight of law behind me).

Perhaps I'll grant you permission to make as many for-profit-copies of
my poems as you like in a book that also contains some of >>your<<
value-added poems, >>your<< art, >>your<< creative contributions and
intellectual property, but only subject to the condition that >>I<< get
to make copies of your work and sell them just as freely (and with as
little return compensation) as you are selling copies of mine.  Perhaps
I might further insist that anybody else who meets this condition can
ALSO republish BOTH of our works on the same basis.

This is hardly "communism".  It is barter, perhaps, but it is an
undeniably >>fair<< barter.  If you choose not to pay this price, write
your own damn poetry and publish away and sell your book for as much as
the market will bear, but leave my poetry out or risk my righteous wrath
etc.  Even if it were ruinously and horribly UNfair, though, it is
scrupulously legal and overwhelmingly ethical, as it defends the very
NON-socialistic notion of the right of an individual to own something,
even something as abstract as "intellectual property".  What in the
world do you think the word ownership is, if not the right and ability
to dispose of and regulate the use of the commodity in question?

This is >>especially<< so in the world of intellectual property.  Is
there anything in the world more absurd than the image of you
"embedding" my poetry in a work of your own, copyrighting that work and
offering it for sale, and then turning around and suing me for using my
own poem because it was copyrighted as a part of your work?  This is one
thing that anyone who has ever written and contributed a GPL package
seeks to avoid.  It is the >>danger<< of offering up intellectual
property "for free" without some toothy licensing conditions.  And yet
it is so simple, and legal, to do, especially in a world where legal
recourse is typically expensive and difficult.  Take a perfectly good
and trivial mathematical result like that underlying the RSA (published
a hundred years ago) and wrap it up and patent it, and suddenly the
descendents of the mathematicians who did the work are paying to use
their ancestor's uncompensated and >>truly<< original work, long held in
the public domain.  Sure it's absurd.  It's also legal, as long as a
court can be made to say so.

Whether or not to use a GPL or LGPL in any given project, especially a
project such as a subroutine library, is a perfectly reasonable subject
for debate.  For what it is worth, I could easily see the GSL offered
under the less restrictive LGPL rather than the stricter GPL, just as a
number of other high level libraries are, to encourage a MIXED use in
full open source project and in commercial products.  If this were
proposed and this were a democracy, I'd support it.  

However, it isn't really a democracy -- the primary authors of the
collective code base called the Gnu Scientific Library have every right
to GPL it, LGPL it, not license it out at all, license it commercially,
or require all of us users to rub blue mud in our navels and pose for a
picture upside down to be sent to them before using it.  They probably
cannot require us to be naked in the pictures -- that far the law might
extend (or at least conflict with other laws and make for a great
argument).

So you might stop sounding like somebody HAS to relicense the GSL so you
can use it commercially in a closed source project and get a bit more
practical.  And no, you can't publish my poems bound in leather and
resell them at $500 a copy (without asking me first).  

Instead, ask nicely -- could we convert the license to LGPL pretty
please?  If that doesn't work, try offering money:  If I use this code
in something I want to keep closed source, I contractually promise to
donate 10% of my net profit to the FSF, for example.  The "owners" of
the GSL can do anything they like, including granting you a specific
exception to their general license (provided that they can somehow ALL
be contacted in a collective and effective way so they can agree and so
forth).  If THAT doesn't work, you STILL have a couple of options.

The fairest and cheapest one is to take advantage of the fact that
NOTHING in the GSL is in ANY way closed.  The algorithms are all open,
uncopyrighted, unpatented, openly published, and public domain.  You can
read any routine, convert it into metacode, rewrite the metacode into C,
or fortran, or basic, or java, and poof -- you have the use of the
algorithm, but not the original code.  So it takes you a few hours of
extra work per routine used -- you'll be paid for that time one day if
your product sells.  Also, do you really need ALL of those routines?
Most applications will only need a handful -- a few days work to port.
Do you really want to use the GSL API for e.g. vectors and so forth?  It
is portable, but not necessarily speed-optimal.  For example, on most
machines it is much faster to generate a whole vector of random numbers
or bessel functions all at once; the latter is already supported in the
GSL, but I'm not so sure about the former.  Rewriting the code will let
you optimize and customize for your own application, although you WILL
need to be SCRUPULOUS in your development process (use CVS! document!)
to avoid a legal suit.

Another one is to use one of the many "competitors" of the GSL.  NAG,
IMSL, Numerical Recipes -- they would be HAPPY to SELL you a numerical
library you can link to your hearts content to your commercial product,
paying them fair market value (a firstborn child, a kidney, a chunk of
your leg) in the best of capitalistic traditions for the privilege.
Just investigating this possibility might give you some idea of the
VALUE of the thing you are insisting needs to be given to you for free
so you can resell it for money.  NR, for example, has a truly laughable
license.  The rest are more straightforward in their usoriousness.

Finally, there is the idea of >>not selling software at all<<.  The GPL
gives you the perfect right to use its code any way you like in a tool
you don't resell and don't publish.  So write your supertool or
superprogram or killer app, and DON'T SELL IT.  Sell the use of it.
Legal as can be.  You can hide the code under a rock, keep it in a
vault, encrypt it so it only decrypts when you use it, and apply it, for
money, to anybody's data via any interface you like as long as you don't
resell or distribute the binaries that include GPL code (as I read the
GPL, at least).  

The GPL only requires that distributed copies of the binary be
accompanied by the right to see the sources, so don't distribute the
binaries.  The web makes this sort of business delivery model easy,
although you'll face the same marketing problem faced by all the
webvendors of services -- finding clients and getting them to web you
work.  The GPL would probably even permit you to LEASE OUT machines that
you own with software you own to companies that rent the use of those
machines without requiring that you disclose the sources of the binaries
on those machines, although that enters into the territory only sanely
trod by an army of lawyers...(does this constitute "distribution of a
binary"?).

I personally am simply grateful that I can use a tool so powerful and
well conceived for free at all, just as I am any time I use any other
feature or tool of Gnu-ware or linux, and have no interest in using the
GSL in any ways where the GPL is a problem in any event.

   rgb

Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy dot duke dot edu