Wrong link...here ya go:

http://groups.google.com/groups/heyletsmakeagame

Indeed.

In fact, reading over my post, I didn't say it quite right.  What I
meant to say, mainly, is that it's very risky to join such a project
without there being any kind of product.  After all, someone could join
and find out later that: (a) the project leader can't write code after
all; (b) the code is horrendously bad; (c) they disagree with the
project leader about what constitutes good software; etc.  If this
happens, then someone has wasted their time in producing whatever
they've done to date.  This then has the potential to get messy.

The flip side of the issue is equally problematic, if not more so.  
There's no code yet, so the leader has to accept people as part of the
project merely because they say that they are.  Functional free software
projects don't work that way.  They operate the simple principle that if
you want to be a developer, you should first write code.  If you
consistently write good code, you are given more responsibility.  This
protects the project, and ensures that high quality code remains the
norm.  It also provides a measure of protection against a certain kind
of developer (a distressingly common one) who requires more of other
people's time than what they actually contribute by being afraid to
write anything without first holding a week-long mailing list discussion
about how to do it.  Without having code, there's a risk of accumulating
developers (and perhaps even a project leader) who aren't actually
contributing, but continue to suck away time and resources by endlessly
discussing things.