[okfn-discuss] incubator.okfn.org

[Jo Walsh]

dear all,

I'd like to add some more reflections based on the experience of OSGeo
and its incubation committee. OSGeo is primarily a software
foundation, but there are some parallels.

Incubation is managed by a committee which has at least one
representative from every current project.  A new project comes along,
the maintainer fills out a comprehensive questionnaire.
http://www.osgeo.org/incubator/process/application.html

Aim of this is to let the committee decide whether the project will
eventually meet the criteria for *graduation* from the incubator. So
the project must already be reasonably healthy, have a group of
committers, ideally some from outside of the organisation that started
the project.
It should be *on track* to establish good governance (not a benevolent
dictatorship),
have the beginnings of a project infrastructure (open repository, bug
tracker, lists etc).
And these are the criteria a project has to meet to achieve graduation.
http://www.osgeo.org/incubator/process/project_graduation_checklist.html

While in the incubator the project will have a mentor - the mentor
should be someone involved with an existing project, one that doesn't
overlap too much in topic area, implementation language, etc. A person
can only mentor one project at a time; there is a limit of about 10
projects that can be in the incubator at any given time. What happens
next?

Graduating from the incubator means a seal of approval - of quality
and reliability of code, reasonable assurance that it really is free
and open, and some guarantee of the future persistence of the project
beyond the original authors - all the things that business users of
free software are looking for.

The incubation committee makes recommendations to the Board both to
incubate projects and to graduate them - I can't remember ever having
seen one sent back in 3 years on the Board.
So there's a high level of trust/delegation involved (and there was a
lot of overlap between the original active board members and the
original incubation committee).

The process *is* pretty selective. Very young, prototype projects end
up in a much more informal "OSGeo Labs" vehicle -
http://wiki.osgeo.org/wiki/OSGeo_Labs - this has a much lower barrier
to entry, self-assessment, and is a way of registering interest in
eventually incubating, and of gathering people to the effort - more
serious than a sourceforge project, but not serious enough yet to deal
with the overhead of incubation.

To me, these two things are at the heart of the incubator:
* Quality assurance
* Mentorship

There's more process overhead than OKF is likely to need - but that
process is meant to help guarantee quality in the work and reliability
for end-users. There's as much focus on getting projects *out* of the
incubator as getting them in.

What would it mean for OKF for a project to have left the incubator -
not that it is "finished" (is anything digital ever finished?) but
that it is vouched for - that to say "This is an OKFN project" is to
be said with pride - it is a mark of quality and of persistence.

I have the idea that there are quite a few funds, labs, foundations
etc aiming to start projects up, create prototypes and provide
bootstrap infrastructure. If we look at the OSGeo model, an OKF
incubator could be a place for those projects to go, and have a stable
home, once the principle is proved.

On Thu, Dec 31, 2009 at 11:27 AM, Michael Fourman
<michael.fourman at ed.ac.uk> wrote:
> Professor Michael Fourman FBCS CITP
> Director, iDEA lab

Michael, I'm just down the street from you, working at EDINA, and I
would love to meet up for a coffee sometime.

cheers all,


jo
--