Subject: Re: A few thoughts.
From: hecker@netscape.com (Frank Hecker)
Date: Thu, 13 Aug 1998 19:56:20 -0400

David Welton wrote:
> If someone comes up with something truly new, even without a patent,
> they can hide their code and have an advantage for a while, being
> the only provider of the code.  If no one else has an implementation
> of it, then why make it open source?  It seems it would make sense
> to do this only after others have released similiar products, and
> the value of having the only implementation has been greatly
> reduced.  Netscape is a fairly decent example of this.

Your comments above are in line with Eric Raymond's argument at
<http://www.opensource.org/secrets.html>; for example, "Closed software
and trade secrets make short-term sense in areas where serious research
is required.... In general, the optimum strategy will be to stay closed
until it becomes clear that someone else is likely to soon go open with
a similar technique, and jump first."

I agree with this argument to some extent; where I mainly depart from it
is in the assumption that this is the _only_ situation where it makes
business sense to keep software closed.  (I also think that the "optimum
strategy" is easier to state than to execute.)  But this is getting away
from your point about innovation and closed source...

> When I talk about innovation, I mean things that are really new, not
> just redoing something else (unless you do it *significantly* better
> - head and shoulders above everyone else).

This is an important distinction that I was going to make in response to
your original message, but you have already done it for me.  It leads in
turn to an interesting question: Under this definition, which of today's
"name brand" open source products represent actual "from scratch"
innovation?  Such software products as Linux, Gnu Emacs, GCC/G++, GNOME,
Apache, and Mozilla seem rather to be (re)implementations of existing
product categories, namely Unix-like operating systems, text editors,
C/C++ compilers, desktop managers, web servers, and web browsers
respectively.

> Some big institutions have the ability to release truly innovative
> things under free licenses (BSD is a good example of this, I think -
> I was rather young at the time and don't know to what degree it was
> a break from the past)

The innovative things you seem to be thinking of here are the products
of research projects funded by institutions whose purpose it is to do
research, not to make money from the software itself.  At least, that
was the traditional situation; it's gotten more complex in recent years
as many universities have tried to retain more control over the
commercial benefits flowing from their research.

This includes imposing more restrictive licensing on
university-developed software, including differentiating in the license
between noncommercial use (permitted) and commercial use (not permitted
without additional licensing).  For example, this is what the University
of Illinois did with the NCSA Mosaic code.  Such "split" licensing of
course means that the software in question is no longer open source
according to the prevailing definition, even though it may be openly
available and incorporates no trade secrets.

So the example of BSD, etc., does not invalidate your argument that an
entity (institution or individual) _wishing to profit directly from
creating innovative software_ is motivated to keep that software closed
in some manner.

It may be more correct to think of innovative software (as a category,
not as a particular product) as having a life cycle as follows:

1. The software category starts out as a research project, with the
particular implementation being either open source or at least with
source available for public inspection.

(Incidentally, we need a simple term to describe software for which
source is published but under more restrictive licenses that don't meet
the Open Source Definition; there is a potential ambiguity here around
the term "open source" which could get as messy as the libre-gratis
ambiguity around the word "free".  Now that I think of it, maybe
"published source" could serve this purpose, by analogy with traditional
published books where the content is publicly available but its use is
subject to legal restrictions.)

2. The software category becomes commercialized, either as closed-source
reimplementations of the original concepts (with possible trade secrets)
or as commercializations of the original software under restricted
(i.e., non-open-source) licensing arrangements.

3. The software category becomes commoditized, and strong open-source
products become available and popular as alternatives to closed-source
products.  The open-source products are in turn reimplementations of
proprietary products (e.g., Linux vs. commercial Unix implementations),
descendants of the original research projects (e.g., Apache vs. NCSA
httpd), or (more recently) "freed" versions of closed-source products
(e.g., Mozilla vs. Netscape Navigator).

Frank
-- 
Frank Hecker          Pre-sales support, Netscape government sales
hecker@netscape.com   http://people.netscape.com/hecker/