just do something already
Along the lines of Joel’s Fire and Motion article, the guys at 37signals are on the right track:
If you find yourself talking more than walking, shut up, cut the vision in half, and launch it. You can always fill in the gaps later. In fact, you’ll know more about what gaps need to be filled after you’ve launched “half a feature” than if you tried to fill them in before launching anything.
Too much thinking can lead to analysis paralysis. Just start the ball rolling, get forward motion and momentum on your project. Obviously some degree of planning is required, but a customer won’t pay you for pages and pages of design documents and fancy flow diagrams. You get paid for a working product. Frequent iterations, tweaking along the way will get you where you need to go. Mind you, this applies well to web applications and smaller teams projects. I could never apply this to a larger project like at my work. The teams are much larger, the release cycles need to be longer and the product is much more complex. Planning is required, but we tend to keep it to a minimum where possible.