Subject: Re: [may be junkmail -pobox] Re: open source definition
From: Keith Bostic <bostic@bsdi.com>
Date: Wed, 22 Apr 1998 14:21:15 -0400 (EDT)

> From: kragen@pobox.com (Kragen)
>
> On Wed, 22 Apr 1998, Jonathan S. Shapiro wrote:
>> The issue of source availability and support is not impeded by whether
>> the original vendor makes a profit.  
>
> I should point out that Cygnus, which has written much of the GNU
> toolset, is making a handsome profit;

I don't believe you've refuted JS Shapiro's argument -- he states that
"the Bazaar model works only when there is an artifact to start from and
react to".  Someone else created something that Cygnus could use.  Had
they needed to create gcc etc. from scratch, their initial funding model
would have been much different.

> RedHat, which has written and
> GPLed a significant part of what's on the RedHat CDs, is profitable
> (although I don't know how profitable; it doesn't appear to be publicly
> held); Crynwr Software, which wrote and GPLed a number of packet
> drivers a while back, made a nice profit on those -- I'm not sure what
> their current status is; Cyclic Software, which includes some of the
> most active developers of CVS, is quite profitable (and has their
> yearly reports published on their web site).  Cobalt Microserver, which
> employs one of the primary authors of the Linux TCP/IP code, looks
> promising.

When this argument was floating a month or so ago, I challenged anyone to
name a company that was writing software and then giving it away -- a
company that sold software for a living, that is, not a company that sells
support and packaging.  Nobody answered that challenge.  To my knowledge,
there is not one single software company that plays the free software game.

>> As I have previously pointed out in this list, and everyone eventually
>> agreed, this model precludes getting early-stage investment to build
>> the initial product with.  
>
> I don't think everyone eventually agreed.  I heard some rather heated
> disagreement from Mike Tiemann at Cygnus, I believe, and I don't think
> anyone actually said, "Yes, I think you're right, John."  Maybe you can
> forward me some email I forgot.

My recollection is the same as Shapiro's -- certainly nobody was able to
come up with an example of where anyone got early-stage investment to
build a software product and then gave it away.

>> That's fine when someone else is giving the software away for you to
>> package, but kindly be honest enough to admit that your scheme only
>> works because you have thousands of unpaid laborers.
>
> Two prongs:
>
> One. Cygnus does the majority of development on the GNU toolchain.
> RedHat has done the majority of development on major parts of Linux.
> So it appears you are mistaken.

At one time, the only thing that made the scheme work was that people had
created the product for RedHat and Cygnus to package/support.  Once they
start providing that packaging/support, yes, it's possible to redirect
money into the development chain.  So, I think you are both partially
correct.  Regardless, the initial development can't be funded using a free
software model.

Keith Bostic				bostic@bostic.com