At the outset of a project nobody knows or understands your ideas and they have no obligations to pay attention to you. You may want to advertise yourself as you begin, or you may want to work alone for some time before deciding if the project is worth continuing.
If you are making a game similar to an existing game you will get much more attention at this stage than if it is attempting to pioneer new concepts. Again, this is because with a new kind of game the ideas won't reach your users until there is something to play.
Project artifacts like documentation don't matter very much at this point. They will only matter if you have users, and if you have no users then you should focus your efforts on getting some.
Early users are a valuable resource. Stay in close contact with them - you want as much feedback as possible. This is where free/open models begin to pay off, since some users may be interested in contributing. Your goals will now start to split between improving the collaboration process and making refinements to the game itself.
Collaborating at an early stage will change the nature of the game significantly, and usually for the better, as it is likely to broaden the game's reach outside of your personal tastes and preferences.
You can use services like Google Alerts to be notified when people blog about your game or ask questions in forums about it. Depending on the nature of your project a lot of people might never bother to contact you directly, so this is an important thing to stay in contact with these users.
Building a Team
Over time some contributors will stay for the long-term. Nobody can be expected to work on a project forever, but if lots of people are involved, it is important to establish an unambiguous model of governance.
Probably the most popular model is Benevolent Dictator for Life, where the central authority comes from one person and decisions are delegated downwards as needed.
In the long term a project may run out of obvious features to implement or new gameplay to add. Gradually interest ramps down. If you want to continue working with the concepts, a fork or "reboot" of the game with new goals may be necessary to revitalize the community.
A mature project's community tends to start meta-discussions about itself. This can be dangerous as such discussions are endless and may subsume original goals.