Subject: Re: A few here may have an opinion on this
From: DV Henkel-Wallace <>
Date: Mon, 28 Oct 2002 17:29:40 -0800

On Monday, Oct 28, 2002, at 14:27 US/Pacific, 
> The real issue is how best to create value with
> software.

This is why the ontological discussion as to "what is a free software 
business?", and various degrees thereof, is so pointless.  I have 
always considered this the mailing list for people interested in 
business issues around the use of libre software, not "free software 

Many of you have heard me say this over the years: "Free software" is a 
how, not a what.  You can no more be a "free software business" then 
you can be a "C-language business".[*]  You can be a business 
developing accounting software, or OSes, or eyeglass frames, but that's 
a what, not a how.  Your business practices, culture, expertise, or 
whatnot has to be domain-specific.

What is interesting is how companies can use (exploit) free software in 
their business strategies.  Thus Red Hat, and IBM follow the same FS 
strategy, differing only in degree.  Likewise Amazon, TiVO and the 
Burlington coat factory.  O'Reilly and SAMS.

A taxonomy of such FSB practices _would_ be a worthwhile discussion.  
Personally I ignore with impatience any discussion of degree of purity.

So all of this _does_ bear on the amount of value -- it should lead to 
greater profits and hence more taxes, else classical economics would 
tell us that the participants wouldn't use FS.

I'd also be interested in a discussion of the "prisoner's dilemma" 
aspects of such businesses: when should you share and when not?  For 
example at Cygnus it was a tenet of (most of[**]) our business that we 
tried to get customer changes merged into the standard tree; when that 
became hard because of politics we formed a new "standard tree".  The 
sales guys never understood it!  "Why not keep our changes 
proprietary?" they'd frequently ask (we could have done with Kerberos, 
for example, but weren't interested in doing so).

But consider someone using BSD software: is it worth sharing up to a 
point, then suddenly stopping?  Could the dread microsoft even use this 
approach (share code under a libre license for several iterations to 
squeeze out a competing standard, then take one implementation 
proprietary and lever their platform dominance to squeeze everyone 
else?).  They kinda tried it with CIFS...but it's a potential 
refinement of IE, that they could do with SOAP.  What kind of defences 
can we really mount to that kind of thing?

Chris sounds heretical, but he's actually on the right track.  Perhaps 
our discussion can become more productive.


[*] Sure, there are companies that only consult on free software, just 
as there are companies that only consult in (say) Perl.  But they are 
at best the exceptions that prove the rule, being, as was endlessly 
discussed, lifestyle businesses.  There are (perhaps) a couple of 
companies like Cygnus and Red Hat that managed to get to significant 
size through heroic effort, but the IRR is inferior to many other 
businesses, and it's not clear to what degree they are generalisable, 
or even sustainable.

[**] I tried hard, and failed, to get Cygnus to develop proprietary 
code to sit alongside our libre offerings.  Despite what I said above, 
I'm not sure in retrospect I was pushing for the right thing.