From: Adam L. Beberg (email@example.com)
Date: Tue Mar 28 2000 - 00:42:13 PST
"Adam L. Beberg" <firstname.lastname@example.org> writes:
> > > Free software enables programmers to deliver much greater value per
> > > hour of work. Partly as a result, Cygnus's income per programmer is
> > > considerably greater than Keane's.
> > Sure it is, all that free help goes right to the bottom line :)
Karl Anderson wrote:
> It helps that Zope is very much a platform, not an app - it gives
> the developer a place to add lots of logic and interface with many
> other components, so a Zope solution usually involves development to
> do complex things.
Cygnus and Zope are not in the same category.
A database (Oracle), a compiler (Cygnus) or a word processor (MS-Word)
emerge out of the box as a working product that do a very well defined
set of fixed functions, usually an RFC/ANSI spec. A database grocks SQL,
a compiler eats source code, a word processor lets you type and use
pretty fonts. They come with a binary that does those set functions and
huge quantities of documentation translated from pig Latin. As a rule
they are either A) no fun to develop with relatively little creativity
because it was all done by 1973, or B) only comprehendable and
improveable by people with PhDs and 10 years of experience.
Documentation is no fun to write and users are extremely costly support.
Everyone has a relative that's a user and no fun to help all the time.
Generally releasing source for these kinds of things would mean they
would never get done to enterprise standards for lack of profit
motivation. It's no wonder the opensource knockoffs of these types of
apps just aren't up to par. Declaring these types of things "opensource"
after the fact is generally a way to get users to maintain the product
when it no longer justifies the costs of having paid developers. One of
todays slashdot articles brings up the "why are all the good databases
commercial" with the usual bias:
Then their is the group where public source make sense. Something like
Perl, or Python, or Zope, or Cosm. The product doesn't come in a box,
and the defined function is to let a developer do what they want with
it. They give you an API/platform to do what you need to do, and
probably a bunch of stock utility apps. Odds are it doesn't do anything
at all without a skilled operator, and the endusers and moms don't even
know they exist at all. The documentation is on the API's for the
platform, not for the apps - that's the developers job. Making the
source public here makes alot of sense, because the code isn't the
product, the API/platform is. Not having source public generally means
the platform/API would never get off the ground to begin with. Everyone
wins when the platform is bugfree, forkfree, faster, and ported to more
stuff. Most everyone that uses it is a developer who is turning around
and selling it to an enduser (apps) or another developer (the platform
or libraries). If they don't start out commercial, they generally have
to go that direction to reach the enterprise level of reliability and
user support. Would Linux ever have gotten on the users desktop without
There is lots of money to be made in both of these realms, but they are
not the same thing. I get just as tired of the "opensource everything"
argument as the "closed source is the only way" one. Each has it's place
depending on many variables, and neither category is an absolute [as my
examples we're always a crisply 0 or 1]. Extremists in both camps are
generally recognized as nuts, but they are needed so we see that neither
works in all cases and most things end up somewhere sane. More and more
things are being done in that fuzzy middle of the real world - with
opensource IPOs, dual licensing, and re-release under quasiopensource.
- Adam L. Beberg
The Cosm Project - http://cosm.mithral.com/
email@example.com - http://www.iit.edu/~beberg/
This archive was generated by hypermail 2b29 : Tue Mar 28 2000 - 00:45:13 PST