[OSGeo-Discuss] OSGeo barriers to entry

[Miles Fidelman]

Landon Blake wrote:
> Andrea Aime Wrote: 
>> I believe the idea that OSGEO projects contributor tend to be paid
>> to work on the project itself is the result the very selection
>> criteria to become an OSGEO project:
>> - mature project
>> - established user base
>> - a formal governance model
> You wrote: "This tells me the project has lots of contributors, lots of
> people that have a stake on it, a big enough user base that the
> possibility of funding is no more a pipe dream but a solid reality.
> Such a project by its very nature will tend to attract more people that
> can find funding to work on the project itself."
>
> I'm curious about how we get a project to the point you describe. That
> seems to be an even greater challenge.
>   
History has some good lessons to teach:

i. A lot of these projects start with a strong motivation and financial 
backing - be it a funded r&d project, a piece of code written for 
internal purposes, or what have you.  Like any development project, good 
design and good project management count for a lot.

ii. At some point, someone makes a decision to make code available.  
This may be informal (a researcher putting some code on their web site 
for people to play with), or more formal (a group or company making a 
deliberate decision to open source some code - for marketing reasons, to 
try to recruit additional development/support at low/no cost).

iii.  A critical step has to happen organically:  People have to 
download and start using the software, and enough people have to find it 
useful that a critical mass develops.  Until this point is reached, 
there's not a lot of need or point to talking about governance models 
(except, perhaps, in making a decision about what license to release the 
code under, and in providing some visibility to get people interested in 
playing with the code)

iv. Once (if) a critical mass emerges, the issue of governance pretty 
much forces its way onto the field.  People start asking "who's behind 
this code? is it stable? is it mature? is it supported?" and so forth - 
and the initial development group is well advised to think long and hard 
about things like: who holds the copyright (does someone need to 
incorporate a non-profit, will the Apache foundation adopt the 
software), where to locate the CVS tree, governance structure (both for 
technical and policy decisions), and so forth.


Seems to me that the keys to getting to the point Andrea describes are:

1. someone has to write some code and then make it available and visible

2. a critical mass of people have to find it useful

3. either the initial authors, or a group of motivated users, have to 
build an initial ecosystem (CVS site, email lists, etc.) - I mention 
"motivated users" because quite a few open source projects are forks of 
other projects (Apache is a fork of the NCSW daemon, the various BSD and 
Linux distributions, etc.)


Having said this, there are times when this isn't quite as organic.  For 
example:

1. a government agency issues a solicitation to develop a piece of code, 
attaching some conditions associated with release and support as open 
source (one of our current projects falls in this category)

2. a group of users/developers sets out to work jointly to develop a 
mutually needed piece of code (though I can probably name more failures 
of this model than successes - IMHO all-volunteer efforts tend to die 
pretty quickly unless one or two people are really driving things, and 
have a pressing need to do so - otherwise, higher priority activities 
tend to pull people away)

3. a commercial firm sets out to develop code on a hybrid 
open-source/commercial business model - there are some good examples of 
this (Aptana comes to mind), but I've seen a lot more of these start 
with some university developed code (Xen comes to mind)


Where OSGeo is concerned: It certainly seems that several projects have 
reached this point (GeoServer, PostGIS for example).  I guess it would 
be interesting to figure out how many users each have, who has 
commercial businesses that are dependent on them, and so forth.  Those 
would be obvious targets for recruiting both funding and people to flesh 
out the developer base (akin to IBM putting a lot of support behind 
Apache - for the simple reason that  their Websphere product line is 
based on Apache).

Miles

-- 
Miles R. Fidelman, Director of Government Programs
Traverse Technologies 
145 Tremont Street, 3rd Floor
Boston, MA  02111
mfidelman at traversetechnologies.com
857-362-8314
www.traversetechnologies.com