Subject: Re: Opportunity lost? Challenge declined!?
From: "Tim O'Reilly" <tim@oreilly.com>
Date: Sat, 12 May 2001 09:23:15 -0700



Russell Nelson wrote:
> 
> Stephen J. Turnbull writes:
>  > >>>>> "Tim" == Tim O'Reilly <tim@oreilly.com> writes:
>  >
>  >     Tim> I have to say I agree very much with what you've said here.
>  >     Tim> An "all or nothing" approach just hardens positions.  I tend
>  >     Tim> to see lots of signs of people coming towards the free
>  >     Tim> software world,
>  >
> 
> I don't believe this to be true.  The benefits derive from the freedom
> to fork.  Remove that -- and all proprietary code removes the freedom
> to fork -- and you remove the benefits as well.

Some of the benefits.  Not "the benefits."  

For example, Java does not allow forking, yet it does allow source
inspection, and that has been extremely helpful to the Java community. 
And yet Java is more successful in terms of number of developers than
any open source language, even Perl.  

Even when forking is allowed, it doesn't happen all that often, because
of cultural norms and community pressures rather than license issues in
any case.  When ActiveState forked perl for the windows version, the
community got together (I admit with some push from me) to heal the
rift.  Sendmail forked into several competing versions, but they were
brought back together, in part by pressure and licensing changes by Eric
Allman.

There are enormous success stories based on forking--the evolution of
Apache from NCSA being the most significant--but they tend to
demonstrate that the importance of forking is primarily in cases where
projects are abandoned or taken in some other direction.  It's a
critical safety valve and a protection that is absolutely critical to
open source.  I don't disagree with that.  But to argue that it's
critical to all the benefits of open source seems to me to be completely
over the top.

In terms of open source, the principles that seem important to me
include:

* Access to source, both for visibility and local bug fixes

* An architecture of participation.  That is, the system should allow
more than a small group of core developers to extend the system, and
should partition it into projects that can be managed by a small core of
developers.  UNIX has this kind of architecture; Windows does not.  As a
result, the original UNIX had all the features of open source without
having a fully open source license.  The loopholes in the license
allowed ATT to kill the goose that was laying the golden egg, but they
didn't stop the goose from laying eggs as long as the farmer wasn't too
greedy.  This seems to me more important than any license term.  

One reason that Linux, Apache, and Perl have all flourished as open
source projects (while Mozilla has struggled) is that they came to
embody an architecture of participation.  The layered architecture of
UNIX/Linux and the pipe/filter paradigm of cooperating tools makes it
easy for anyone to add a tool without having to get the permission of
someone else.  The module structure of perl and apache allows developers
to  take the core software as a platform and extend it in directions of
their own without having to get the permission of the maintainers of the
core.  

You can even argue that the fundamental architecture of the web, with
things like CGI, the separation of presentation logic from content, and
so on, is what makes it possible for a kind of "open source" community
to flourish on top of servers that need not themselves be open source. 
I'm with Mitch Kapor:  "Architecture is politics."  Lessig expounds on
this point at great length in his book Code.

This is certainly true of the internet.  Like UNIX, it has an
architecture of participation.  As long as you adhere to the protocols,
the endpoints can do whatever they want.  

* Documentation that matches the modularity of the software.  The man
page plays a larger role in open source success than many people credit.

* Openness to a community for comment and feedback, and all the
principles Eric Raymond lays out in the Cathedral and the Bazaar,
including social norms that encourage people to contribute, that reward
them with reputation for doing so

License terms like the right to fork, and the right to redistribute
under the same terms, are *protections* of open source effectiveness,
not causes of them.

Lao Tzu says:

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

The laws or licenses we create are needed because people have lost sight
of or never understood why open source works (the way of life, the
science and market dynamics of why this is an effective software
development paradigm), or they have lost goodness, and look to subvert
the system for short term local benefit.  They don't drive the system.


> 
> Very simple test case: BSDI.  Shared source business model, long
> before Microsoft ever "invented" the idea.  And they even did it
> right, by allowing BSDi customers to share source with each other.
> So, how successful was BSDi over Walnut Creek, which served as the
> commercial publisher of FreeBSD.  Or Wasabi Systems, which is a NetBSD
> publisher.

You tell me.  I don't know enough about the axes of success you're
considering to even know how to answer that question.  Some possible
axes are wide adoption, financial success, reputation in the community,
etc.

If, for example you consider financial success:

Very simple test case:  Microsoft.  Proprietary source wins hands down
on financial success.

If you consider wide adoption and social impact:

Very simple test case:  The WWW.  Public domain, with no protections,
and multiple proprietary forks, has had wider adoption than any flagship
open source project, and in fact is a superset of many of the most
successful ones.

For that matter, it's a horse race between open (the internet) and
proprietary (Windows) about which one has had a bigger social impact. 
And in fact, Windows has had wider adoption.  And contrary to the
opinions often offered on this list, there are probably as more
developers on Windows than there are on UNIX/Linux, and many of them are
just as passionate about their system as we are about UNIX/Linux.

If you consider reputation:  Bill Gates, Tim Berners-Lee, and Linus
Torvalds all have pretty elevated status, so reputation might just be
orthogonal to licensing.  That is, free software licensing may be a
technique that allows a hacker to build reputation, but so may building
a successful proprietary software business.

But I'd be interested in any event to understand the data behind your
assertion. How do *you* measure the relative success of these three BSD
variants.  Sales figures?  Number of users?  etc.



> 
>  >     Tim> We need to encourage signs that people are moving in the
>  >     Tim> right direction, towards openness, even if they aren't all
>  >     Tim> the way there.
>  >
>  > True, there is a need.  But should serving that need be in the charter of
>  > this list?  Especially when their goal is almost certainly not "free,"
>  > and never will be?  I think that the problem of running a business
>  > based on free (by the strict definition) software is interesting,
>  > important, and relevant to enough developers to deserve its own list.
>  > That's what FSB is, right?
> 
> That's always been the idea.  You can see the first message to the
> list by sending mail to fsb-get#0@crynwr.com.
> 
>  > We need a separate list for the purpose of encouraging movement toward
>  > openness by people who are starting from proprietary.  I wish the term
>  > "open source" hadn't been coopted as a marketing ploy.  It's the
>  > perfect name: "Open Source Software Business."  Maybe "Published Source
>  > Software Business" is a good enough alternative?
> 
> Why not "Shared Source Software Business"?  You'd get a lot of
> subscribers @microsoft.com.
> 

Personally, I think it's an extremely interesting challenge, and one
with a lot of social benefit (which is my main goal) to get Microsoft to
open up and realize the benefits of cooperation as well as the benefits
of competition, and to find a better balance between the two.  Just as I
urged Allchin and Mundie and other MS folks to talk to open source
folks, I urge you guys to be more open to talking to Microsoft.

Cass Sunstein's book republic.com talks about the dynamics of groups
that talk only to people who already agree with them:  they tend to
become more extreme, reinforcing their existing beliefs, while groups
that regularly engage in dialogue with people who have opposite views
tend to moderate towards the middle.  

Now, I have nothing against this group wanting to sit in a corner and be
extremists, if that's what you want.  But since this is the premier
gathering and conversation place for people who really care about the
intersection of free software or open source and the business world, I'd
much rather see us put our collective weight into engaging with "the
enemy".  Yes, it might cause those of us in dialogue to become more
moderate, but my firm belief is that it will help Microsoft to become
more moderate as well.
> --
> -russ nelson will be speaking at http://www.osdn.com/conferences/handhelds/
> Crynwr sells support for free software  | PGPok | Mailing lists should not set
> 521 Pleasant Valley Rd. | +1 315 268 1925 voice | Reply-To: back to the list!
> Potsdam, NY 13676-3213  | +1 315 268 9201 FAX  | http://russnelson.com/rt.html

-- 
Tim O'Reilly @ O'Reilly & Associates, Inc.
101 Morris Street, Sebastopol, CA 95472
+1 707-829-0515, FAX +1 707-829-0104
tim@oreilly.com, http://www.oreilly.com