I find it helpful to maintain a closure ratio; I shoot for perhaps 1 out of 5 projects that get past one night's worth of fiddling should reach "closed", for some definition of "closed". It's crippling to think that you should finish everything you start. It's better to always start things and never finish than never start anything at all, but it's better to "finish" every so often too. This gives me the satisfaction of occasionally finishing things, while still letting me wander broadly. It's a powerful combo, depth and breadth.
I do this too. I manage it by periodically reviewing my open projects and either pruning or building a kind of mission statement–go or no go. If they make the next step it doesn't mean they have to get to the end, but it means you should spend some time scoping and figuring out what completion might eventually mean.