Re: Open source WO (was: Apple, Microsoft, Java and the Web?)

On Aug 24, 2006, at 10:26 AM, [email protected] wrote:

> Also, as I said a few days ago in the context of a GVC Sitemaker  
> discussion, it's a very risky thing for a company to open-source a  
> product they depend on themselves.  They have to have resources  
> available to monitor and control what happens to that source;  they  
> have to maintain a high level of quality and keep things moving in  
> a direction that is compatible with their own needs.

Since this line of reasoning is being used more and more often, I  
guess some debunking is due ;-)

This fear comes from not understanding how open source works. If a  
company doesn't want open development, just opening the code is a  
pretty minor benefit to the users. So why bother? Just to claim "open  
source-ness" for marketing purposes?

On the other hand if the company does it to grow the community around  
the code, then it has to invest into building the community, and give  
control to the community (otherwise this is just a gimmick). If this  
is the case, the company must understand the dynamics involved and  
implement procedures that would make the project flourish. A few  
points to note:

* Projects that were taken over and messed up by strangers usually  
had all signs of abandonware before that happened. I.e. the owners  
didn't care.

* You don't have to give commit access to total strangers. Volunteer  
developers need to understand that they have to demonstrate the  
ability to fit in the specific community before they are granted  
commit rights. This means they'll have to submit their contributions  
via patches for as long as it takes to build trust.

* All development should be done in public. I.e. if I am about to  
blow a subproject from the repo or rename all classes to follow a new  
naming convention that I just invented, a proper notice to the  
mailing list is due. And others have the right to veto the change (if  
they can explain why).

* There is source control to do conflict resolution after the fact.  
 From experience rolling back is rarely needed, except for the most  
trivial cases of human errors.

Try it, it works :-)

Andrus