I like the approach that "thegamedesigner" does
(The project Broken Brothers)
Which was made for the experimental gameplay competition.
http://experimentalgameplay.com/blog/

This approach is basically a rapid prototyping and
polishing of an unusual gameplayidea or Artstyle
in a limited time (7 days)

The good thing with this approach for indie development is, that
you then dont try to make the gretest game ever, but
try to make a unique idea or use an unique Artstyle
wich can be nevertheless produced quickly.

I would suggest these small closed projects for
indie developers, rather that trying to make
a huge project with several members.
Simply because those large projects drain the exitement
and interest away, once you are over the prototyping stage
and go into the slavework of creating content, models and levels.

My main motivation for the project I did was to create
a prototype of some never or seldom used technologie of
"big" games
by using gamestudio. (and Java)
including FPS Ai, RTS, different Mulitplayer approaches,
shadow rendering etc.
They usually dont result in shipable projects, but
train a lot in understanding the development approches.
The fun is to reaseach the topics, have ideas and
make them work in code soemhow.

(I worked on commercial Games in several companies,
where this knowledge was a good kickstart for getting the job)

Bottom line: this engine and the Indie development should
be seen as a hobby and exercise platform.
Its not bad not to finish all dreamprojects.

You should either try to make a small unique game and
finish that, or try to create the next WoW-Age-of-Counterstrike
wich will fail, but teaches a lot in creating parts
of it.