Subject: Re: Open Source Developer (Economics)
From: Ian Lance Taylor <ian@airs.com>
Date: 21 Mar 2003 14:53:18 -0800

Ken Dyke <ken_i_m@elegantinnovations.net> writes:

> On Thu, 2003-03-20 at 10:56, Ian Lance Taylor wrote:
> > * The design decisions to be made are largely technical decisions
> >   which can be easily made by the developers.  For large scale
> >   business software, the design decisions must be made by people with
> >   experience in large scale business.
> > 
> > Siebel requires
> > extensive customization for each deployment, 
> 
> I think it safe to say that most FSB developers come from a *nix
> background.  A fundamental *nix design principle is to provide
> "mechanism" and to push "policy" as close to the end user as possible. 
> Those coming from a Windows background generally do not understand this
> and whine about the over-whelming number of choices they have to make
> just to get things running.  This delay in setting policy is a feature
> not a bug.
> 
> "extensive customization for each deployment" is just another way of
> saying "setting policy".

True.  However, while there is a similarity between the Unix design
principle and something like the Siebel approach, I think it would be
misleading to push that similarity too far.  Siebel requires extensive
customization because it is purchased by large businesses who already
have their own ways of doing things, and changing their existing
processes to match the software would be a lot more expensive than
changing the software to match their processes.

> Claims that the open source development model are not up to the
> complexity of this problem space smack of the same claims that were
> leveled against Linux and the complexity of operating systems.

True.  Note that I did not make that claim.  I did make the claim that
building software to be used by people working in large businesses
require extensive input from people who have worked at the appropriate
position in large businesses.  I make the further claim that very few
software developers, particularly free software developers, have that
experience.

> As a community of FOSS developers emerges that is focused on the
> enterprise level/problem space the dynamics of this development model
> will come into play.  Then either SAP, Siebel, Oracle, et al adapt or
> get buried.  (Or hire a bunch of lobbyists of buy off Congress a la
> Hollywood.)

It could certainly happen.  On the other hand, we shouldn't make the
mistake of assuming that every problem space is going to attract the
attention of free software developers.  It might happen, or it might
not.

Ian