The hardest part for me is getting the ball rolling. My primary method is to pick the most fun thing I can think of to do. I try not to feel guilty for starting with something easy or fun, because it's such a good way to get going. Once I'm on a roll it's not nearly as difficult to tackle the nastier problems.
Isn't that a way to fall behind on schedule and implementing things you don't need? I guess you're assuming that what you can pick from is what you HAVE to do.
I think the idea is that you make a point not to let the idea of drudgery get you down early-on, because it can be overwhelming to the point where you get nothing done.
In the context of an afternoon's work, it might mean you start with a small utility that you know is good but might not have any immediate value, or some optimization tweak that's been itching you for awhile. That'll get you immersed into the project and excited about further possibilities, so tackling drudge-work won't seem so bad.
That idea also applies more generally, and is often the idea alluded to when you hear people around here talk about banging out a 'Version 1' as fast as possible.
