Subject: Re: Sun, BSD, and GNU
From: "Tim O'Reilly" <tim@oreilly.com>
Date: Mon, 31 May 1999 09:24:48 -0700



Craig Brozefsky wrote:
> 
> "Tim O'Reilly" <tim@oreilly.com> writes:
> 
> Leveraged it into Free Software, or leveraged it into proprietary
> software?  Which license lets companies leverage free software into
> proprietary code with the most profit return, is not a question I
> think is worth pursuing, particularly in a "Free Software Business"
> context, since by definitions such companies would not be Free
> Software companies.

OK, if that's the question (and if everyone agrees it is), I will
abdicate my argument.  I *thought* the original thread was
sparked by the question of "which license is best for business"
not "which license is best for the creation of more free
software".  Since the latter is the primary aim of the GPL, it is
certainly more likely to hit that target.

My experience of the university licenses (which is the generic
term I'd prefer to use for the Berkeley/X11 family of licenses),
on the other hand, is that their aim has simply been the
advancement of the art, the enrichment of the soil of the
computer industry, from which further free OR commercial efforts
can grow.  My assessment is that they are more likely to hit
*that* target.

Perhaps a better question than the one originally posed is "which
license is more likely to spur innovation and the creation of
added value?"  I would argue that most of the really innovative
libre software out there is from the university tradition, while
most GPL software is a politically-motivated reimplementation of
work that has already been done.

Them's fighting words, I know, but I think there is some truth to
them.  The basic aim of the GPL is to promote the use of libre
software, and as a result, the primary focus has been on
replacing proprietary software.  Since university-style licenses
don't have that goal, they tend to focus on doing work that
hasn't already been done.

That isn't to say that now that Linux is in vogue, and there's a
business model to back it up, the GPL isn't also being used for
innovative work. It is. There's all kinds of innovative stuff
happening in the GPL world today that wasn't happening a few
years ago, when there wasn't the commercial interest.  (For
instance, I was pleasantly struck recently by a conversation with
a database researcher at IBM, who said that Linux has an enormous
amount to spur good database research in the past year or so. 
This has been driven by the need to support scalable web-based
databases.)
> 
> What is the "paid in dollars" total for businesses which have produced
> *Free Software* from BSD code?
> 
> What is the "paid in dollars" total for businesses which have produced
> *Free Software* from GPL code?

If that's your measure (i.e. dollars don't really matter, only
dollars generated while creating more libre code), then I would
have to agree that the GPL is quite clearly the best license,
since it's the only one with the explicit goal of creating more
libre software.  But I thought we already knew that.  I'm
certainly not arguing that point.  If I wanted to make sure that
no one produced a proprietary version of my product, I'd use the
GPL.  But if I didn't care whether someone built something
proprietary on top of my work, but just that they added value,
I'd use a university-style license.  

To my mind, the benefit of a university-style license is that it
says:  "if you don't add value, this pretty much has to be gratis
(because the libre/gratis version is also available), but if you
can add enough value that people are willing to pay you for it,
more power to you."  This encourages people to add to the
software who are not just volunteers.

As I've always said, the choice of the right license depends on
the aims of the software's creator.  Both GPL and
university-style licenses are pretty good at hitting their aims;
they just have different aims.

The final point (and then, I really am going to quit this
discussion), is that this whole licensing arena is really in flux
right now, because the economic landscape of the software
industry is changing so much.  The GPL, via Linux, is definitely
undermining the ability of people to sell proprietary software,
and changing the business models to ones that treat software as a
commodity, with the business value created not by the software,
per se, but through added value such as branding or channels of
distribution.

Whether this is ultimately a good thing for the software industry
is yet to be seen.

I'd say it's likely to be bad for the software industry as we
know it--or should I say, "used to know it", since the software
industry has pretty much imploded anyway due to Microsoft's
predatory practices.  Software is going to be a low-margin
industry, just like PC hardware.

So one question is where people are going to be making money in
this future we're building.  I see three or four places:

* Hardware manufacturers, who will save on their windows
licenses.

* Companies who sell a mix of free and proprietary software, a la
IBM with Apache and WebSphere.  (An interesting side note on
WebSphere.  It is a suite of proprietary applications that work
with Apache, but interestingly enough, I believe it is largely
"gratis" (though not libre), bundled into some humongous IBM
service offerings rather than "sold" as a product.)

* Commodity distributors of software, a la Red Hat (the Dell of
the operating system layer)

* Builders of new *services* on top of the commodity software
layer.

To me, this last bullet is the BIG point that everyone is
missing.  Craig said earlier that we shouldn't consider companies
that just *use* libre software, and don't produce it, in
considering the economic impact of free software.  But I consider
that the real "wave front" of innovation in the computer industry
right now is precisely among those companies.  ISPs, information
providers like Yahoo!, ecommerce companies like Amazon.  They are
building on the open source platform, and not returning a whole
lot to it.  But you guys don't rag on them, because they don't
distribute software; they just provide services, and so you don't
think of them as software businesses.

Amazon uses all kinds of GPL'd software in their business, but no
one is saying "hey, you have to open source the code for that
associates program" because it doesn't look like "software" to
you guys.  No one is saying to Earthlink, "hey, you hacked
sendmail to support 1,000,000+ users, so you need to give it to
your competitors as well."

I'm not suggesting that you should rag on these companies...but I
do believe that if you did, they'd stop using GPL'd software to
the extent they do, but would still be able to use software
created under the university-style licenses which are more
friendly to the idea of creating added value and charging for it.

(This is a side-point, but because the GPL focusses on software
distribution, and because software distribution as we knew it in
the PC era is becoming a smaller and smaller part of the way
software is being delivered today, it may end up becoming less
and less relevant.  No one is thinking about what licenses we
need for the future shape of the industry.)

Overall, to return to my theme, people like to say that
university-style licenses have been abused, as when companies
take the main line of code development private.  But even when
this happens, nothing is *taken away* from the community. 
Further value is not given, but the libre community is still free
to continue its own work, if it so desires.

What's more, I think that the behavior of companies (like Sun)
who did this in the past is likely to be different today.  Then,
the dominant idea was that proprietary was absolutely key to the
success of a business.  Now, there is a much more flexible idea
that there are useful business tradeoffs between  libre/gratis
and proprietary.

The ultimate argument for university-style licenses is precisely
that freedom is a good thing.  As businesses realize the
benefits, coercive licenses *may* be less necessary.  As Lao Tzu
says,

   Losing the way of life, men rely on goodness.
   Losing goodness, men rely on laws.

You could turn this around and say something like:

   Men rely on laws to cause others to do what is right.
   Men rely on goodness to cause themselves to do what is right.
   The ideal is so to know it is right that you do it like
breathing.

What we need to do is to create more awareness of the benefits of
libre access to source code, and the importance of giving back to
enrich the soil of innovation.  This is the "science" of libre
software--the idea that it *works better* as a software
development methodology, and as a way of giving users the ability
to solve their own problems.

Still, it can take a while for this realization to sink in.  This
is one reason I'm starting to turn the focus of my evangelism
towards companies like Yahoo! and Amazon, urging them to figure
out what, from their archives of software they've built to run
their businesses, they can contribute back to the soil of the
open source movement.  Because they don't distribute software, no
current software distribution license touches them.  But I think
that for their own ultimate good, they want a large part of the
software they develop to be out there in the world as well as
inside their companies.


-- 
Tim O'Reilly @ O'Reilly & Associates, Inc.  
101 Morris Street, Sebastopol, CA 95472   
707-829-0515 ext 266, Fax 707-829-0104, tim@ora.com 
http://www.oreilly.com http://software.oreilly.com
http://www.songline.com