Subject: Re: Definition of open source
From: Rick Moen <>
Date: Sat, 6 Nov 2004 15:31:01 -0800

Quoting Alan Rihm (

> Hello,

Greetings and salutations.  Let's get one piece of dumb rhetoric out of
the way immediately, because I see you've got Marius all excited over it:

> Why should someone be able to make money on someone else's code,
> without any financial responsibility to the originator of that code?

By default, they should not.  Fortunately, copyright law prevents this 
economic disaster from happening, so we can all heave a gusty sigh of
relief that your right to issue instances codebases under proprietary
terms is _not_, in fact, being ripped away from you.

A celebration party to give thanks for OSI not being a worldwide, cruel
dictatorship over all software authors is hereby left as an options for
any readers so inclined.

Moving right along:

> Currently the number one requirement from the OSI to be considered
> "open source" is "Free Redistribution". It seems to me that this
> requirement is not by itself the key point to determining if a project
> is open source or not. 

Well, OSI clearly does not agree.  Citing reasons should not be
necessary after the long history of this matter, but I touch on it
below, anyway.

> My perception is that this requirement means that the only new
> projects that can be justified are those originated by developers who
> wish to further their resume, or companies with deep pockets who feel
> they can out-code the competition. Of course the exception to the rule
> are those who were early entrants, and have time and wide-scale
> adoption on their side.

Actually {checking for a moment}, many of the several thousand software
packages on the Linux and BSD workstations and servers in my household
seem to be exceptions to that -- except you've cleverly left "furthering
their resume" as a catch-all.  I'm not going to try to convince you that
you can make money from open source:  Maybe your situation and
particulars make attempt that a bad idea.  Others are either doing so or
finding it to have compelling advantages for other reasons -- e.g., use
value being enhanced by consequences of the development model.  Some
business models for open source are described in essays linked from here:

But anyhow, you describe this as a "problem":  OK, but problem for whom?  
My sad lack of a summer cottage in Langue d'Oc is a "problem" from my
own idiosyncratic perspective, but not one I expect license-discuss to
assist me with.

If I understand correctly, the "problem" you see is your inability to
find a way to hand out copies of a licence under OSD-compliant terms
without recipients having the right to freely redistribute that
software, creating a low-or-no-cost secondary market that potentially 
impairs your control of the product.  

Well, I'm willing to go to the extent of extending sympathy for your
dilemma, but your suggestion that OSD be edited for your business
convenience is a non-starter.  Either don't use an OSI-certified
licence, or find something other than a tollgate on redistribution that
offers value, e.g., trademark-based branding, proprietary rights
available only for money on a second, differently-licensed code
instance, support contracts, upgrades, etc.

> The other problem that I see, is that the "dual license" strategy causes
> people to launch a project, and then essentially end-of-life the project
> to encourage people to buy the paid/closed version.

Really?  How many years has Aladdin Ghostscript (recently renamed to
AFPL Ghostscript) been a counter-example, guys?  I can't quite remember.
Hmm, looks like v. 1.0 was released Aug. 11, 1988.  Sixteen years, then.

> As for distribution, let each project owner carve out what can and can't
> be done. This would foster so many more projects, and it would
> ultimately help end users and developers.

It's useful, here, to haul out one of the traditional gedankenexperiments,
that of Prof. Daniel J. Bernstein's (eventual) demise.  (I should stress
that I wish Prof. Bernstein a very long and happy life -- as well as any
shot at immortality he can wrangle.)  

Prof. Bernstein has generously issued several highly regarded codebases
he has authored over the years (qmail, djbdns, ezmlm, and others) with
(only) limited rights of redistribution.  Specifically, the right to
distribute derivative works was never granted.  The code availability
has made a number of devoted users happy, but it's precisely the
"ultimately" aspect that's problematic, because only Bernstein has the
right to develop and release new versions.  At his demise, absent new
provisions to the contrary, those codebases will become for all
practical purposes unmaintainable -- for lack of the right to
distribute derivative works.

Now, you say OSI should "let each project owner carve out what can and
can't be done":  This is either suspiciously overblown special-pleading 
rhetoric on your part, or you are deeply confused, because OSI _does_
"let" people do that, in the sense that it has no say whatsoever over
what coders may permit or not with their codebase instances.  

But, if there are the tollgates on redistribution that you so
desparately want, then it's not open source, and cannot get OSI

> After all, aren't we really trying to foster options to closed source
> solutions, and give the end-users and developers freedom to make
> software work the way they want it to work?

Personally (and speaking only for myself), I don't give a tinker's damn
if there's yet more source-available shareware, which is what you, like
Marius, are talking about.  Been there, done that.  But feel free to
promote it, since you like it.

Cheers,                 There are 10 kinds of people in the world, those who 
Rick Moen               know ternary, those who don't, and those who are now     looking for their dictionaries.  -- Ron Fabre