OSI Certification Process

[Alex Nicolaou]

Bruce Perens wrote:
> From: Alex Nicolaou <anicolao at cgl.uwaterloo.ca>
> > Besides, I have discovered that amongst my users and potential contributors
> > there exists a great apathy about what license I use anyway!
> 
> In that case, I urge you to withdraw your request for Open Source
> certification of the SOS license. It would do active harm to the free
> software community for OSI to accept yet another license that contributes
> as little as this one over the existing licenses. The only contribution
> you're making a real case for is simplicity. Unfortunately, that's
> deceptive: no attorney was involved in the license creation and there
> is on assurance that unsimple implications could be lurking unseen in
> your terms.

In the particular case I'm currently working on I have to discuss this
with my co-author. We had both agreed that we felt it was important that
our license got OSI certification, so that there would be no future
confusion over whether the code is really "Open Source" or not. We felt
that getting OSI certified was a way for us to make it easier for
commercial distributions to include our software.

In addition, from afar the OSI certification seemed simple enough: read
the OSD, read the other licenses, construct a license that met the
criteria and subject it to public review. 

I've attempted to do all of these things with an open mind, and have
tried especially hard to be receptive to commentary and always get
specific commentary that enables me to change the license so that it is
more satisfactory.

I think it is very important to add the above additional criterion to
the WWW pages, if the opinion you express above is part of the
certification process. The idea that a license should be considered not
suitable for certification solely because it "contributes little" and
that certifying a new license can "do active harm to the free software
community" is not at all reflected on the WWW pages. If this info had
been on the web site, it possibly could have saved me a lot of time and
effort. 

As to the simplicity not being real, I don't really know what to say
there. The license is clearly a simple one. Its language is
straightforward. In the Canadian legal system, this type of license
would be more binding and usable in court because it would be clearer if
the person violating the license knew that they were doing so during the
violation. I realize that the legal system in the United States is much
tied up with having everything written in a very specific style, and
maybe this makes my license a poor one for use in the USA; this question
can only be answered by review from a US attorney.

I do agree that simplicity is the main goal of the SOS license. However,
the remaining goals of formalizing the distribution and update process,
providing a framework where the question of "what is a derived work" is
irrelevant, and explicitely permitting patch distribution under totally
different terms also have value in my mind. 

In closing I should say that I think it is a mistake to withhold OSI
certification simply on the basis of a licence's merits, and not on some
more rational basis such as whether it meets the OSD or not. To do so
will not stop the flood of new licenses; it will only embitter new
entrants to open source development and either turn them away from it
altogether or lead to them ignoring the OSI certification issue and
simply saying "this is Open Source software".

The vast majority of software produced does a job that another package
already does. It's no wonder that the people who produce all of this
redundant software are likely to produce redundant licenses as well.
Spreading FUD by withholding OSI certification from these licenses is
not the right way to go. One alternative might be to have an OSI Gold
License certification, which says that not only is the included license
OSI certified, but it is recommended for use on new programs. Only
licenses with a long track record (e.g. more than 3 years) and a history
of being used on many programs (e.g. more than 50 programs) would be
eligible for OSI Gold status. Then you get to allow new entrants to the
field while still making it clear which licenses are the preferred ones.
By having objective criteria for moving a license from OSI to OSI Gold
status you make it clear that it is an impartial distinction, created by
the overall success of the license amongst developers.

alex