Something I tried to write about in my article in LJ, is, yes, every 
project needs to start somewhere. But you need to evaluate your own 
capacity, and compare that against the maturity of the software and the 
community. You need more internal capacity to deal with immature 
software with an immature community. Now, if there weren't those who had 
that that willingness to take risks, had that capacity, and were willing 
to apply it to a new project -- then we'd never get mature projects, 
indeed. But it's not necessarily the right decision for everyone.  It 
requires capacity and a willingness to be a developer, not just an end 
user.  Which is of course a spectrum, and ideally no user of 
collaborative open source is 'just' an end user, but some are more 
towards the 'developer' end than others.

You need to evaluate your own capacity and measure it against where the 
project you are considering is at.  No individual software is right or 
wrong for everyone universally, it depends on the context.

I tried to write about some of this stuff at more length (among other 
issues) in this article:


Thomas Krichel wrote:
>   Brett Bonfield writes
>> I think Jonathan and Nicole nailed it with community health,
>   I beg to differ.
>   If you requiree a healthy community to start working with a piece of
>   software, how do you want a grassroots project to start? Obviously a
>   small project will start with one or two developers, and it won't
>   grow, until a few people work with it despite the fact that it's a
>   small thing to start with.
>   Requiring an upfront healthy community is particurly problematic is
>   a small community such as digital library work.
>   On the other kind, there is widely adopted software that I got
>   cajoled into maintaining, that consider bad. Apache is one of
>   them. I run maybe 50 virtual servers an a bunch of boxes, I am still
>   puzzled how it works and it's trial and error with each software
>   upgrade, where goes that NameVirtualServer thing into, the constant
>   croaks "server foo has no virtualserver". I'm not a dunce, but
>   Apache makes me feel I am one. When I look at these config files
>   that are half-baked XML, I wonder what weed the guy smoked who
>   invented this.
>   If I could do it allover again, I would do it in lighttpd. Oh well
>   it was not there in 1995 where I started running web servers. 
>   Other problematic case: Mailman. I run about 130 mailing lists, over
>   80 have a non-standard config, I am running every few months into
>   problems with onne of them, despite the fact that I wrote a script
>   to configure all the non-standard lists the same way.
>   Cheers,
>   Thomas Krichel          
>                                                skype: thomaskrichel